DOCUMENT RESUME CE 050 798 Hamilton, Gayle; And Others · DOCUMENT RESUME. CE 050 798. Hamilton,...
Transcript of DOCUMENT RESUME CE 050 798 Hamilton, Gayle; And Others · DOCUMENT RESUME. CE 050 798. Hamilton,...
ED 298 285
AUTHORTITLE
INSTITUTION
SPONS AGENCY
PUB DATENOTEAVAILABLE FROM
PUB TYPE
EDRS PRICEDESCRIPTORS
IDENTIFIERS
ABSTRACT
DOCUMENT RESUME
CE 050 798
Hamilton, Gayle; And OthersSaturation Work Initiative Model in San Diego.Interim Report.Manpower Demonstration Research Corp., New York,N.Y.
California State Dept. of Social Services,Sacramento.Aug 88362p.
Manpower Demonstration Research Corporation, ThreePark Avenue, New York, NY 10016.Reports Research/Technical (143)
MFO1 /PC15 Plus Postage.* Demonstration Programs; *Incentives; *Job SearchMethods; Program Development; Program Effectiveness;Program Implementation; *Referral; *WelfareRecipients; *Work Experience ProgramsAid to Families with Dependent Children; California(San Diego); *Saturation Work Program; Work IncentiveProgram
Tne San Diego (California) Saturation Work InitiativeModel (SWIM) was part of the Social Security Administration'stwo-site demonstration of saturation work programs in an urban area.The mul'Licomponent program operated from July 1985 until late 1987,when it was replaced by a new statewide initiative called the GreaterAvenues for Independence (GAIN) program. The model was designed totest the feasibility of requiring the continuous participation ofwelfare applicants and recipients in a series of work-relatedactivities for as long as they remained on welfare. The programactivities included job search, work experience, and referral toeducation and training programs. The program was examined from thestandpoint of two perspectives--the extent to which the full WorkIncentive (WIN)-mandatory caseload was involved in employment-relatedactivities at a particular point in time and the coatinuity ofindividuals' participation in the program. During typical months ofthe demonstration, approximately half the WIN-mandatory caseloadsubject to the participation requirement was active in one of theprogram's components. Although many program eligibles did notparticipate on a continuous basis, it was concluded that San Diegoreached the maximum feasible participation rates for the type ofprogram in question. (MN)
XXXXXXxxXxxxXXxxXXXXxxxXxXxXx******************************xxxXxxxxxxx** Reproductions supplied by EDRS are the best that can be made *
X from the original document. *xxxxXXxxXXxxXXxx**xxxXxxxXxxxxXxx**x**xxxxxxxxxxxxXxxxxxXxXXxxx*mmxxxx*
I NTERM REPORT ON THE
Gayle-natnilton U.S DEPARTMENT OF EOUCATIONOffice of ducational Resetra and Improvement
EDUCATIONAL RESOURCES INFORMAnONCENTER (ERIC)
his document has been reproduced asreceived from the person or organizationoriginating it
C Minor changes have been made to improvereproduction Quality
Points of view or opinions stated in this docu-ment do not necessarily represent OfficialOERI positiOn or policy
"PERMISSION TO REPRODUCE THISMATERIAL HAS BEEN GRANTED BY
Manpower DemonstrationResearch Corporation
Augtist 1988
TO T"C EDUCATIONAL RESOURCESINFORMATION CENTER (ERIC)."
BEST COPY AVAILABLE
r
BOARD OF DIRECTORSRICHARD P. NATHAN, ChairmanProfessorWoodrow Wilson School of
Public and International AffairsPrinceton University
M. CARL HOLMAN, Vice-ChairmanPresidentNational Urban Coalition
PAUL H. O'NEILL, TreasurerChairman and CEOAluminum Company of America
ELI GINZBERG, Chairman EmeritusDirectorConservation of Human ResourcesColumbia University
BERNARD E. ANDERSONManaging PartnerUrban Affairs Partnership Corporation
RAMON C. CORTINESSuperintendentSan Francisco Unified School District
ALAN KISTLERPresidentHuman Resource Development InstituteAFL-CIO
ELEANOR HOLMES NORTONProfessorGeorgetown University Law Center
ISABEL V. SAWHILLSenior FellowThe Urban Institute
ROBERT SOLOWInstitute ProfessorMassachusetts Institute of Technology
GILBERT STEINERSenior FellowBrookings Institution
MITCHELL SVIRIDOFFProfessor of Urban PolicyNew School for Social Research
WILLIAM S. WOODSIDEChairmanExecutive Committee of the Board of DirectorsPrimerica Corporation
JUDITH M. GUERONPresidentManpower Demonstration Research Corporation
MDRC
INTERIM REPORTON THE
SATURATION WORK INITIATIVE MODELIN SAN DIEGO
Gayle Hamilton
with
Vilma OrtizBarbara GoldmanRudd Kier stead
Electra Taylor
!..anpower Demonstration
Research Corporation
August 1988
lf.
This evaluation was funded under a contract fromthe California State Department of SocialServices. The findings and conclusions in thisreport do not necessarily represent the officialposition or policy of the funder.
Copyright 1988 by the Manpower Demonstration Research Corporation
5
ACKNOWLEDGMENTS
While the authors share responsibility for this report, they benefitedfrom the contributions of many individuals in California and at MDRC.
As principal investigator for the SWIM evaluation, Gayle Hamilton hadthe lead role in all work in San Diego and in writing this report. BarbaraGoldman was responsible for the evaluation design and guided the researcheffort since its inception. She also played a major role in shaping thisreport. Vilma Ortiz conducted the estimation of program impacts and wrotethe impact chapter. Electra Taylor oversaw data quality assessment acti-vities as well as the noncompliance study, and wrote the research designchapter. Rudd Kierstead produced the participation estimates, assisted inthe analysis and handled special studies of childcare and nonparticipation.
Many facets of the research were aided by the continued support ofstaff in the County of San Diego and State of California. In San Diego,the study benefited from the cooperation of a number of key people. JohnRobbins, then head of the SWIM Division within the Department of SocialServices (DSS), consistently supported the research objectives by providingaccess to data and to staff involved in various facets of the SWIM program.Critical assistance also came from the staff of the SWIM Division, parti-cularly Chris Morrison and Cesar Rivas; Joan Zinser, then chief of thecounty's Workfare Division, and her staff, espeCially Tom Walters; RayKoenig, then head of the county's Employment Preparation Division, alongwith his staff, particularly Lee Lovell and Caron Johnson; and Steve Olsenfrom the Employment Development Department (EDD) District Office of SanDiego County, along with EDD staff. In addition, Mildred Mura of the DSSIncome Maintenance Division and John McCann of the DSS Data ProcessingDivision provided access to valuable data. The strong backing of TedSchwend, director of the San Diego County DSS Employment Services Bureau,was critical. MDRC was also fortunate to have the support of RandallBacon, then director of San Diego County DSS.
In DSS at the state level, Steve Munro, the project's contractofficer, gave continuing assistance throughout the implementation of thestudy. In EDD at the state level, Dale Shinn and Tad Funakoshi wereresponsible for providing MDRC with the earnings data files.
In addition, the authors were fortunate to have input from individualsat the U.S. Department of Health and Human Services. In particular, HowardRolston, currently associate administrator with the Family SupportAdministration's Office of Policy, provided critical and insightfulcomments on the design, implementation and analysis of the study.
At MDRC, numerous staff made important contributions to the project.Judith Gueron, president of MDRC, and Michael Bangser provided valuablecomments on earlier drafts of the report, most particularly the conclusionsdrawn in the Executive Summary. Fred Doolittle, Jim Riccio and KaySherwood comprehensively reviewed the entire report. In addition, Kay
6
Sherwood coordinated MDRC's work with California during the design phase ofthe project. Support was also provided by Sharon Rowser, who interviewedprogram staff in connection with education and training activities. In the
Information Services Department, Karen Paget supervised the processing ofthe data files, aided by Darlene Hasselbring, who was instrumental in
designing and monitoring San Diego's automated tracking system and randomassignment procedures. Within this department, Anita Kraus and ShirleyJames provided assistance. In the Research Department, Daniel Friedlanderwas an advisor on the impact analysis; Gregory Hoerz played a key
coordination and advisory role in a variety of areas; John Wooden producedand helped analyze the impact estima;:es; Mark Levinson provided programmingassistance; and Sara Cohen and Elizabeth Eisner, assisted by Steve Walsh,coordinated report and table production. Felicity Skidmore edited thereport with the help of Carla Fine, Alex Kopelman and Robin Freedman. PattPontevolpe, Claudette Edwards and Stephanie Amy Cowell contributed theirword processing skills.
The authors also benefited from comments on design documents and anearlier draft of this report provided by members of MDRC's Board of
Directors, Gordon Berlin of The Ford Foundation, and a special AdvisoryCommittee to MDRC's Demonstration of Work/Welfare Initiatives.
-iv- 7
The Authors
PREFACE
This is the first of two reports on a welfare employment initiative
operated in San Diego, California as part of the Social Security
Administration's two-site Demonstration of Saturation Work Programs in an
Urban Area. Philadelphia, Pennsylvania was the other site. The study
looks at a multicomponent program that operated from July 1985 until late
1987, when it was replaced by a new statewide initiative, the Greater
Avenues for Independence (GAIN) program.
The demonstration is of particular importance because of the unusual
nature of the program and the strength of the research design. San Diego
tested the feasibility of requiring continuous participation of welfare
applicants and recipients in a series of work-related activities for as
long as they remained on welfare. The activities included job search, work
experience, and referral to education and training programs. The study was
intended to determine the maximum feasible level of monthly participation,
as well as to provide information on the sensitivity of measured rates to
different definitions of participation. The impacts of such a program on
employment and welfare receipt were also of primary interest.
This report covers issues of implementation and participation, and
presents short-term impacts on employment and welfare receipt. The final
report, scheduled for 1989, will examine longer-term impacts and compare
the program's benefits and costs.
-v-
wilw..-,-_ffix
We hope that the findings fran this evaluation will contribute to
informed decision-making and ultimately lead to the development and
operation of even more effective programs designed to increase the
self-sufficiency of welfare recipients.
Judith M. GueronPresident
9
EXECUTIVE SUMMARY
For approximately two years, starting in July 1985, the County of San
Diego in California operated the Saturation Work Initiative Model (SWIM) as
part of the Social Security Administration's two-site Demonstration of
Saturation Work Programs in an Urban Area. Philadelphia, Pennsylvania was
the other site. The demonstration, directed towards individuals applying
for or receiving benefits under the Aid to Families with Dependent Children
(AFDC) program, was intended to test the feasibility of having at least
three-quarters of program-eligibles active in a welfare employment program
at all times as well as the impact of such a program on employment and
welfare receipt.
The Manpower Demonstration Research Corporation (MDRC) has been
evaluating SWIM under a contract from the California Department of Social
Services. This report, the first of two, examines the feasibility of
continuously serving a large proportion of the WIN-mandatory caseload. In
San Diego, this included both people on AFDC (mostly women who are single
heads-of-household with children aged six or older) and on AFDC-U (mostly
dien in two-parent households). Together, these groups accounted for
approximately 40 percent of the welfare caseload.
The report examines different ways of measuring participation and the
factors that affected participation levels. Short-term impacts on employ-
ment and welfare receipt are also presented. A final report, scheduled for
completion in 1989, will examine longer-term impacts and compare the
program's benefits and costs.
-vii- 10
Policy Significance of SWIM
As part of a broader debate about the conditions that should properly
be attached to the receipt of, welfare, there is strong interest in learning
the extent to which AFDC recipients can be required to participate in
employment-related activities in return for their grant. The SWIM
demonstration grew out of a general philosophy that the participation
requirement should be greater than had been the case in most previous
welfare employment programs, which tended to impose short-term obligations,
often for only a part of the WIN- mandatory caseload.
BV explicitly seeking to maximize the proportion of the entire
WIN-mandatory caseload that participates for the full duration of their
stay on welfare, SWIM provides an opportunity to begin determining
realistic benchmarks for programs which similai:ly try to 'saturate' the
caseload. Development of these benchmarks also bears directly on the
congressional debate about whether -- and, if so, at what level --
participation standards should be set for welfare employment programs.
Participation rates can be examined fran several perspectives. Two of
those addressed in this report are particularly relevant. The first
perspective focuses on the extent to which the full WIN-mandatory caseload
was involved in employment-related activities at a particular point in
time. This is measured as a :monthly participation rate' -- the proportion
of individuals eligible for the program during a month who actually parti-
cipated during that month. The second perspective concerns the continuity
of individuals' participation, and is measured as the percentage of
registrants who participated during every month they were subject to the
program's requirements.
These two measures, together with an examination of the reasons why
individuals did not participate, help define the feasible upper bounds of a
participation (or !saturation:) requirement. As this report will des:rihe,
however, the results are sensitive to the method by which participation
rates are calcul.3ted and to the specific conditions in San Diego during the
pericd of the SWIM demonstration. These and other factors, such as program
costs and benefits, need to be taken into account in assessing the
feasibility of achieving particular participation levels.
Overview of Findings
The SWIM program operated in place of WIN in two of the most urban of
the county's seven welfare administrative areas, comprising about 40
percent of the county's caseload. Program activities included job search
assistance which taught participants has to locate and obtain unsubsidized
jobs; the Employment Work Experience Program (EWEP) which required
recipients to work in public or nonprofit agencies in exchange for their
benefits; and referrals to community education and training programs.
During typical months of the demonstration, approximately one-half of
the WIN-mandatory caseload subject to the participation requirement w--
active in either job search, work experience, education and training, or
employment while still registered with the program. About one-third
participated during almost all the months that they were eligible for the
program.
Despite the fact that many program - eligibles did not participate on a
continuous basis, the report concludes that San Diego reached the maximum
12
feasible participation rates for a program such as SWIM. This is suggested
by a number of factors. Individuals who did not participate in a particu-
lar month were often only temporarily inactive. In addition, most of those
who did not participate had legitimate reasons for being inactive, such as
waiting for program components to begin. In all, only about 10 percent of
those eligible in any month failed to participate without a program-
approved reason.
There are a number of reasons for caution in generalizing the San
Diego findings to other areas or situations. For example, the results were
very sensitive to the types of activities counted as *participation' (e.g.,
program-arranged activities, self-initiated education and training, or
in-program employment) as well as to those who were counted as "program -
eligible.' In addition, the particular conditions in San Diego -- a
relatively good labor market coupled with California's high grant levels,
the county's extensive community education and training resources, its
lengthy prior experience in operating similar employment initiatives, and
the availability of supplementary funding -- suggest that comparable rates
might not be achieved in other areas of the state or nation with less
favorable conditions. An examination of participation levels in similar
saturation programs in other locations is needed to determine the range of
levels achievable under varied conditions and resources.
Finally, calculation of the participation rates presented in this
report hinged upon the existence of comprehensive and accurate data. Much
of the data came from a management information syscem established
specifically for the demonstration. Participation rates could not readily
be calculated for programs,lacking such data systems.
The study's impact analysis showed that, at least in the short-run,
SWIM increased both employment and earnings and reduced welfare receipt for
AFDC registrants (called AFDC-FG's in California). These positive impacts
were sustained throughout the nine- to twelve-month follow-up period and in
sane cases were still increasing at the end of this period. For the AFDC-U
registrants, SWIM increased employment and reduced welfare payments,
although impacts on earnings were less consistent. The employment gains
for the AFDC-U's are of particular interest in light of the absence of
employment impacts in several prior MDRC studies of programs for this
population. The final report on SWIM will extend the analysis to determine
whether the impacts are sustained over a longer period of time.
Program Context
As noted above, San Diego was an unusual setting in which to test the
feasibility of operating a multi-component program with an ongoing parti-
cipation requirement for most of the WIN- mandatory caseload.
First, during the period SWIM operated, the San Diego economy was
relatively healthy. Unemployment rates in the county were below those for
the State of California and the Unit'd States as a whole. Additionally,
program registrants benefited from the tenth highest welfare grant level in
the country. California's high AFDC grant levels, coupled with San Diego's
healthy economy, enabled more program registrants to combine unsubsidized
employment with the receipt of welfare than would be the case in other
areas. In SWIM, employment of at least 15 hours per week fulfilled the
program's participation obligation.
Second, unlike sane areas of the country, San Diego has an extensive
network of educational and training facilities. The availability of these
opportunities increased the likelihood that SWIM registrants could, on
their own initiative, participate in these programs. This network of
services also facilitated SWIM staff's placement of registrants in these
activities.
Third, unlike many welfare agencies, the San Diego County welfare
department had lengthy experience in successfully implementing welfare
employment programs. The SWIM model itself was an expansion of the
county's previous program, which involved three-week job search workshops,
followed by 13 weeks cf work experience. Prior to that, the county had
experimented with workfare programs for food stamp recipients. This
experience reduced some of the start-up issues that might otherwise have
been expected in SWIM and contributed to relatively smooth program
implementation.
Fourth, the county's regular WIN allocation was supplemented by State
Employment Preparation Program (EPP) monies and by special federal
demonstration funds. SWIM, therefore, does not test program participation
rates achievable if only WIN funding were available.
It is also important to understand the characteristics of the individ-
uals who entered the program, since the nature of the targeted population
probably affected participation levels and program impacts. Among the AFDC
registrants in the impact sample, 39 percent were applicants and almost all
were females, with an average age of 34. Forty-two percent were black, 27
percent were white, and 26 percent were Hispanic. Eight percent of the
AFDC sample spoke only Spanish. Fixty-six percent had a high school
diploma or GED. Over half of the sample had received welfare for at least
'...1111611=11111191111iNr
five years during their entire life. Almost half of the sample had never
been employed throughout the two and one-half years prior to initial
program registration.
Among the AFDC-U impact sample members, 60 percent were applicants.
Ninety-one percent were male, averaging 33 years of age. Fortv-two percent
of the AFDC-U's were Hispanic, and 16 percent spoke only Spanish. Twenty-
five percent were white, 20 percent were black and 11 percent were Asian.
Less than half (46 percent) had a high school diploma or GED. The AFDC-U
sample did not have as much of a welfare history as the AFDC sample: Less
than 15 percent had received welfare for at least five years during their
entire life. Less than one-third had never been employed throughout the
two and one-half years prior to registration.
Program Model
The SWIM program model specified set sequences of components, depend-
ing on registrants' activities as of and prior to registration. The
majority of individuals were assigned to an initial two-week job search
workshop. The first week of the workshop consisted of group sessions
designed to build self-confidence and job-seeking skills. In the second
week, registrants used telephone banks to call prospective employers.
Individuals who did not find employment by the end of the workshop were
referred to EWEP, or work experience, in which they were required to hold
positions in public or nonprofit agencies for up to 13 weeks. Concurrent
with EWEP, registrants were referred to biweekly job clubs, which were
usually similar to the 'telephone,' portion of the job search workshops.
Those who completed EWEP and job clubs without finding employment were
assessed to determine their next activity. Options included Adult Basic
Education programs, courses for General Educational Development (GED)
diploma preparation, English a Second Language (ESL) programs, skills
training, on- the -job trainins and additional job search activities. SWIM
did not operate or fund education or training activities. Rather, program
staff referred registrants to existing community programs.
There were several variations to the above sequence, depending on
whether registrants had attended job search workshops or EWEP as part of a
previous program. In addition, registrants who were enrolled in education
or training activities at the time of initial program registration were
allowed to continue in these activities if they met the program's
requirements as to content, duration and intensity. Similarly, registrants
who were employed at least 15 hours per week at the time of initial program
registration were not assigned to program activities. However, once a
registrant completed or dropped out of a self-initiated education or
training program or stopped working, he/she was assigned to the regular
SWIM program sequence.
Implementation Findings
Registrants generally proceeded smoothly through the program,in spite of a complicated case management structure.Substantial staff resources were required, however, to carryout case management tasks.
Several different agencies or staff units provided program services
and served as case ma' tgers to registrants progressing through the SWIM
model. Each set of staff provided one type of program activity. As
17
-xiv-
registrants completed an activity, they were referred to a different set of
staff.
Two factors contributed to the relatively smooth progression of
registrants from one component to another. First, the county had extensive
experience with this type of case management. Second, the fixed-sequence
nature of most of the SWIM model provided staff with clear guidelines on
activity assignments.
Program staff were occupied by case management tasks to a greater
extent than the provision of direct program services. Two-thirds of
professional staff time connected with SWIM was spent on case management
tasks: monitoring attendance, dealing with noncompliance, arranging
support services and tracking registrants' activities. Over one-third of
the staff time spent on case management consisted of arranging and
authorizing support service payments.
Although the SWIM automated tracking system was designed toaid in case management as well as provide data for theresearch, the system functioned primarily as a data depositoryand not as an interactive system that provided timelyassistance to staff in tracking registrants.
In part, the time case managers spent on tracking registrants' acti-
vities reflected the fact that the SWIM automated tracking system was not
fully exploited. One set of staff was responsible, along with their other
program duties, for ensuring that the automated system reflected all
registrant activity. Given the many program components in SWIM, the
various sets of staff involved and the program's ongoing participation
requirement, this task was very time-consuming. .hough the system could
have been used to do much of the clerical work in tracking registrants'
activities, the county did not have the staff or resources to develop
1-XV-
computer routines that would allow local offices to make extensive use of
the system.
Participation Within a 12-Month Follow -Up Period
About two-thirds og those eligible for SWIM services partici-pated in some activity within 12 months of initial programentry. Most participants were active in job search, withfewer in MEP or education/training activities.
Slightly over half -- 51 percent of the AFDC registrants and 57
percent of the AFDC-U registrants -- participated in some type of job
search activity during the follow-up period. Most of these participants
were active in two-week job search workshops, generally the first type of
activity to which registrants were assigned.
Participation was not as common in work experience, which generally
followed job search activities. Approximately 19 percent of both AFDC's
and AFDC-U's participated in EWEP during the 12 months following
registration.
Approximately 24 percent of the AFDC registrants and 17 percent of the
AFDC-U registrants participated in education or training activities within
the follow-up period. The majority of these participants were placed in
these activities by program staff. However, the extent to which SWIM
increased registrants' enrollment in education and training beyond what
individuals would have done on their own is not yet clear. This
Information will be available for the final SWIM report.
According to program activity data, over one-third of the SWIM-
eligibles were employed at some point for at least 15 hours per week while
remaining registered with the program. Individuals who were employed less
19
-xvi-
than 15 hours per week were assigned to regular program activites, while
those employed part-time (15-30 hours per week) were deferred. Individuals
employed more than 30 hours per week were eventually deregistered from the
program.
Eleven percent of the AFDC's and 8 percent of the AFDC-U's were sanc-
tioned for noncompliance within the 12-month follow-up period. These rates
were higher than those observed in most of the programs evaluated as part
of MDRC's Demonstration of State Work/Welfare Initiatives.
Levels of Monthly Participation
As discussed above, several different measures were used to address
the extent to which SWIM imposed an ongoing participation requirement on a
substantial proportion of those eligible for the program. Findings using
two of these measures are outlined below. The first measure en.imates the
percent of those eligible for the program in any given month who partici-
pated during that month (i.e., monthly participation rates) and indicates
the types of activities in which registrants were participating each month.
The second measure examines the extent to which individuals participated
during every month they were eligible for the program, by calculating the
proportion of registrants' program-eligible months in which they
participated.
Approximately 8,300 individuals were eligible for program services
during the two-year time span of the demonstration. Once the existing
caseload was phased into the program, an average of 3,592 individuals were
eligible for services in any month.
20
Monthly participation rates varied greatly, depending on thetypes of activities counted. During the second year of SWIM,average monthly participation rates were 22 percent if onlythose in program- arranged services are included, increased to33 percent when education and training activities initiated byregistrants are also counted, and reached 52 percent byfurther including employment while registered.
Figure 1 presents monthly participation rates for each of the 24
months of the demonstration. Rates for SWIM as an ongoing program are best
reflected in the figures for the second year, when the existing WIN-regis-
trant caseload had been phased into SWIM. In any one month during this
period, between 18 and 28 percent of those eligible at least one day for
the program participated in job search, work experience, or program-
arranged education or training. Defined in this manner, monthly partici-
pation rates were similar for AFDC's and AFDC-Ws.
When the definition of participation was expanded to also include
registrant-initiated education and training, monthly rates were from 31 to
35 percent during the second year. This proportion was similar for AFDC's
and AFDC -U's.
If employment while registered is also counted as participation, the
rates ranged from 47 to 55 percent. Monthly participation rates calculated
in this manner were slightly higher for AFDC's than for AFDC-U's, because
welfare regulations permit AFDC's to combine work and welfare to a greater
extent than AFDC-U's.
Approximately one-third of program-eligibles participatedduring almost all the months they were in the program. Thisincluded individuals who were in th,,: program for relativelylong periods of time.
On average, registrants were eligible for SWIM services for 7.8 months
during the 12-month follow-up period. A small proportion of all regis-
21.
100
bR
0cc
5 50
FIGURE 1
SWIM
PERCENT OF INDIVIDUALS ELIGIBLE FOR SWIM DURING EACH MONTHWHO PARTICIPATED IN PROGRAMARRANGED ACTIVITIES, SELFINITIATED
ACTIVITIES OR EMPLOYMENT. BY TYPE OF ACTIVITY
JUL85
22
OCT85
JAN86
APR JUL66 85
Month
OCT86
JAN87
Employment
;;iSotf-In'rtiatedI : Activities
UProgram-ArrangedActivities
APR87
23
trants 16 percent -- either participated in program activities or were
employed in each month they were eligible.
However, if a few months of inactivity are allowed, 36 percent of
registrants were active in almost all (70 percent) of the months they were
program-eligible. On average, this group was eligible for the program for
8.0 months daring the follow-up pel.od. Continuous participation was as
likely for individuals with long periods of eligibility as for those with
short periods.
While monthly participation rates fell short of the 75 percentgoal, an examination of reasons for nonparticipation indicatedthat San Diego achieved the maximum rates possible in SWIM.
According to case file reviews, close to 90 percent of those eligible
fcr program services in any month were either active or otherwise complied
with program requirements during the month, even if they did not parti-
cipate. Only about one-tenth were inactive due to noncooperation or
program staff failure to assign or follow-up registrants.
Among those inactive, many were assigned to components (e.g., job
search workshops, EREP worksite positions or education/training programs)
scheduled to begin during the next month. Some individuals were temporar-
ily excused from participation due to illness or other situational factors.
Same were pending deregistrations. A small number were undocumented
workers, who were required to register for the program but were not
required to participate.
As this suggests, many of those who did not participate in a given
month were only temporarily inactive. In fact, about one-quarter of those
inactive participated in either the month before or the month following
24
their inactive month. In all, two-thirds of those inactive in any month
participated at some point in SWIM.
Implications for Setting Participation Standards
SKIM shows that participation rates are affected by manyfactors, including the program model and local conditions.Consequently, localities will face unequal challenges intrying to achieve the same monthly participation rates.
The SWIM program model placed sane constraints on measured partici-
pation. First, the evaluation only counted as program participants people
who were active in specific, real components: job search workshops or job
clubs, EWEP, or education and training. Administrators interested in
maximizing measured rates could instead define participation in a less
stringent manner, by also including assessments, assignment to an activity
or participation in less demanding services for example, ongoing
individual job search.
Second, in SWIM, some temporary inactivity occurred when individuals
made the transition from one component to another. While administrative
actions could reduce some of this (e.g., by scheduling job search workshops
more frequently), other lags were not subject to program control (e.g.,
cycles for particular education courses).
The SWIM findings also suggest that the local setting can affect parti-
cipation in a number of ways. For example, in a typical month of the SWIM
demonstration, 19 percent of the WIN- mandatory caseload fulfilled their
participation requirement through employment of at least 15 hours per week.
In other states, jobs might not have been as readily available or
alternatively, part-time work would have effectively moved someone off
welfare. Other factors that can be expected to vary across locations
include the extent of self-initiated education and training, the rates of
welfare turnover, and the characteristics (and employability) of welfare
recipients.
For all of these reasons, localities will face different challenges in
achieving any particular participation rate. Further, this rate may
reflect very divr:se combinations and intensities of activities across
sites.
Although the concept of a monthly participation rate is arelatively simple one, the calculation of such rates requireshigh-quality data.
Both San Diego County and MDRC staff invested substantial time to
ensure that the SWIM automated tracking system contained all data items
necessary for the research. Further, complicated programming was needed to
calculate monthly participation rates. Without this type of data or
programming capability, it would have been difficult to present such a
comprehensive 'f participation. At the very least, time-consuming
manual revie f program case files would have been necessary to measure
participation rates.
In addition, monthly participation rates were very sensitive to the
quality of the data used in the analysis. Due to the Importance of these
data to the research, county staff spent increased Lime on the collection
and monitoring of program tracking data.
Impact Findings
To evaluate the impacts (or program effects) of SWIM, individuals in
the current WIN-ma.m,' 'ry caseload of the two SWIM offices, along with any
26-xxii-
individuals determined to be WIN-mandatcry during the year, were rant.
assigned to an experimental or control group during the first 12 months of
the program. Those in the experimental group were required to participate
in SWIM; members of the control group ere not assigned to SWIM activities
but could, on their own initiative, enroll in community programs.
Impacts were estimated by comparing the welfare and employment experi-
ences of the experimental and control groups over time. All experimentals
-- both participants and nonparticipants as well as those who found employ-
ment and those who did not -- were compared to all controls. Outcome
differences between experimentals and controls were considered statis-
tically significant if there was no more than a 10 percent possibility that
they could have occurred by chance. Follow-up data on welfare payments
were obtained from AFDC records maintained by the County of San Diego;
earnings data came from the California Unemployment Insurance system.
In this report, impacts are examined for a relatively short period
after random assignment -- approximately nine months for employment and
earnings and 12 months for AFDC receipt. The final report will include
longer follow-up.
For AFDC registrants, SWIM resulted in statistically signif i-cant employment and earnings gains as well as welfare savings.
During the full nine-month follow-up period, 46.4 percent of AFDC
experimentals were employed at some point compared to 36.4 percent of the
controls. (See Table 1.) This is a statistically significant difference
of 10 percentage points or an improvement of 27 percent. Over this same
period, experimentals had average earnings of $1,442 compared to $1,185 for
controls. This represents a statistically sign: 'icant earnings gain of
0 ky
TABLE 1
SWIM AFDC SAMPLE: SHORT -TERM IMPACTS ON EMPLOYMENT,
EARNINGS, WELFARE RECEIPT, AND WELFARE PAYMENTS
Outcome and Follow -Up Period Experimentals Controls Difference
Ever Employed, Quarters 2-4 (%) 46.4 36.4 +10.0**
Average Number of Quarters withEmployment, Quarters 2-4 0.97 0.76 +0.22***,
Ever Employed (%)
Quarter of Random Assignment 27.9 25.1 +2.7Quarter 2 30.8 24.6 +6.3Quarter 3 32.9 25.3 +7.6*Quarter 4 33.5 25.7 +7.8
Average Total Earnings,Quarters 2-4 (5) 1442.00 1185.47 +256.54***
Average Total Earnings (S)Quarter of Random Assignment 296.68 285.23 +11.44Quarter 2 371.00 338.L.1 +32.76Quarter 3 497.69 392.03 +105.66Quarter 4 573.31 455.19 +118.12 **
Ever Received Any AFDC Payments,Quarters 2-5 (%) 91.1 91.9 -0.7
Average Number of Months ReceivingAFDC Payments, Quarters 2-5 8.59 9,12 -0.53***
Ever Received Any AFDC Payments (%)Quarter of Random AsSignment 91.2 91.4 -0.3Quarter 2 89.7 89.7 -0.1Quarter 3 78.9 81.5 -2.6**Quarter 4 70.6 76.0 _5.5.Quarter 5 65.8 72.4 -6.5***
Average Total AFDC PaymentsReceived, Quarters 2-5 ($) 4424.00 4827.08 -403.08***
Average AFDC Payments Received ($)Quarter of Random Assignment 1193.27 1194.12 -0.86Quarter 2 1286.17 1331.93 -45.76**Quarter 3 1119.45 1224.55 -105.10***Quarter 4 1031.55 1159.81 -128.27***Quarter 5 986.83 1110.78 -123.95***
Sample Size 1606 1605 3211
NOTES: These data include zero values for sample members not employed andfor sample members net receiving welfare. A twa-tailed t-test was applied todifferences between experimental and control groups. Statistical significancelevels ore indicated as: = 10 percent; = 5 percent; *** = 1 percent.
2 8
TABLE 2
SWIM AFDC-U SAMPLE: SHORT-TERM IMPACTS ON EMPLOYMENT,EARNINGS, WELFARE RECEIPT, AND WELFARE PAYMENTS
Outcome and Follow-Up Period Experimentols Controls Difference
Ever Employed, Quarters 2-4 (%) 53.4 44.0 +9.3***
Average Number of Quarters withEmployment. Quarters 2-4 1.13 0.94 +0.19***
Ever Employed (5)
Quarter of Random Assignment 37.9 35.8 +2.1Quarter 2 36.3 29.2 +7.1***Quarter 3 37.9 31.7 +6.2**Quarter 4 38.9 32.8 +6.0**
Average Total Earnigs,Quorters 2-4 ($) 2364.77 2027.77 +337.01
Average Total Earnings ($)Quarter of Random Assignment 600.73 601.44 -0.71Quarter 2 656.56 564.15 +92.41Quarter 3 837.00 692.33 +1A1.67*Quarter 4 871.21 771.29 +99.93
Ever Received Any AFDC Payments,Quorters 2-5 (%) 86.2 86.4 -0.3
Average Number of Months ReceivingAFDC Payments, Quarters 2-5 7.57 7.93 -0.37
Ever Received Any AFDC Poyments (X)Quarter of Random Assignment 85.6 84.4 +1.2Quarter 2 83.4 83.9 -0.5Quarter 3 67.5 71.0 -3.5Quarter 4 64.7 67.4 -2.7Quarter 5 59.9 62.7 -2.8
Average Total AFDC PaymentsReceived, Quarters 2-5 ($) 4873.97 5298.34 -424.37**
Average AFOC Payments Received (s)Quarter of Random Assignment 1263.87 1274.29 -10.42Quarter 2 1418.75 1470.22 -51.48Quarter 3 1191.54 1321.01 -121.47***Quarter 4 1165.63 1279.25 -113.62**Quarter 5 1098.05 1227.85 -129.800*
Sample Size 687 654 .341
NOTES: These data include zero values for sample members not employed andfor sample members not receiving welfare. A two-tolled t-test was applied todifferences between experimental and control groups. Statistical significancelevels are indicated as: * = 10 percent; ** = 5 percent; *** = 1 percent.
-XXV --4% 2a
$257 -- a 22 percent increase. Earnings gains appear to result more from
changes in the percent employed than from higher wages or more hours worked
among those employed.
For the full 12-month follow-up period, total welfare payments
averaged $4,827 per control and $4,424 per experimental. This is a
statistically significant welfare savings of $403 per experimental -- an 8
percent reduction in welfare payments. Reductions in grant payments
increased through the fourth quarter and then leveled off.
It is also notable that beginning in quarter three, SWIM resulted in
statistically significant reductions in the percent of experimentals who
were still on the welfare rolls. BY the final quarter of follow-up, 65.8
percent of experimentals were receiving welfare as compared to 72,4 percent
of controls, yielding a statistically significant difference of 6.5 percent-
age points. In addition, experimentals received welfare for one-half month
less than controls.
Both AFDC applicants (those who were in the process of applying for
welfare at the time of random assignment) and recipients (those already
receiving welfare at the time of random assignment) experienced employ,-
ment, earnings and welfare Jaapacts.
Based on results for an early-enrolling group of registrants, impacts
for the AFDC group appear tc continue through 18 months of follow-up.
For AFDC -U registrants, SWIM resulted in statisticallysignificant employment gains as well as reductions in welfarepayments. However, increases in earnings and reductions inthe percent receiving welfare were, for the most part, notEtttistically significant.
During the full nine-month follow-up period, 53.4 percent of the
AFDC-U experimentals and 44 percent cf the AFDC-U controls were employed at
-xxvi-
30
some point. (See Table 2.) This is a statistically significant increase of
9.3 percentage points or a 21 percent improvement. These employment gains
were sustained throughout the follow-up period.
Experimentals had average earnings of $2,365 and controls had average
earnings of $2,028 during the full follow-up period, for a statistically
significant difference of $337. This is an earnings gain of 17 percent.
Except for quarter three, the quarterly earnings impacts were not statis-
tically significant.
During the 12-month follow-up period, total welfare payments averaged
$5,298 for controls and $4,974 for experimentals, yielding a statistically
significant reduction in welfare payments of $424. 'Welfare savings
continued throughout the follow-up period. There were, however, no statis-
tically significant reductions in the percent of AFDC-U experirentals
receiving welfare during this period.
Among AFDC-U registrants, both applicants and recipients had employ-
ment gains and reductions in welfare payments. There were earnings gains
among both groups, but these differences were not always statistically
significant, partly due to small sample sizes.
An examination of 18 months of follow-up for an early- enrolling group
of AFDC-U registrants indicated that both employment gains and welfare
savings were sustained over this longer-period of time. There appear to be
earnings increases in the later quarters as well.
Based on a preliminary investigation, it appears that for AFDCapplicants the SWIM impacts are consistent with those reportedfor San Diego's earlier job-search/work experience demonstra-tion. Among AFDC-U applicants, employment gains may be largerfor SWIM than in the prior demonstration.
Preliminary comparisons suggest that SKIM's short-term impacts on AFDC
0 !
applicants are generally similar to the impacts for welfare applicants in
San Diego's earlier job search/work experience program. There is sane
evidence that SWIM's Short-term impacts on the AFDC-U applicants' employ-
ment gains may be larger than those observed for the earlier demonstration,
although welfare impacts are consistent between the two programs. However,
there are a number of factors other than the differences in the two program
models -- such as the health of the economy, the characteristics of clients
and the offices in which the programs were operated -- that should be
considered when comparing these results.
A more complete picture of SWIM's effectiveness will be available in
the final report, to be completed in 1989. It will contain substantially
longer follow-up on employment and welfare receipt. It will also present
information on the extent to which SWIM increased registrants' enrollment
in education and training beyond what individuals would have done on their
own. Finally, the report will examine program costs to determine whether
they were offset by program benefits.
- xxviii-
CONTENTS
PAGEACKNOWLEDGMENTS iiiPREFACE
EXECUTIVE SUMMARY viiLIST OF TABLES xxxLIST OF FIGURES xxxivDEFINITION OF ACRONYMS xxxvi
CHAPTER
1 INTRODUCTION 1
2 RESEARCH DESIGN, SAMPLES AND DATA SCURCES 24
3 ADMINISTRATION AND STAFFING 50
4 THE NATURE OF PROGRAM SERVICES 70
5 OVERVIEW OF REGISTRANT FLOW THROUGH THE PROGRAM 97
6 OTHER PARTICIPATION ISSUES 127
7 MONTHLY PARTICIPATION RATES 143
8 IMPACTS ON EMPLOYMENT, EARNINGSAND WELFARE OUTCOMES 190
APPENDICES
A TECHNICAL APPENDIX: PROGRAM TRACKING DATAUSED IN THE REPORT 224
B SUPPLEMENTAL TABLES TO CHAPTER 2 242
C SUPPLEMENTAL TABLES TO CHAPTER 5 250
D SUPPLEMENTAL TABLES TO CHAPTER 6 252
E SUPPLEMENTAL TABLES TO CHAPTER 7 260
F SUPPLEMENTAL TABLES TO CHAPTER 8 264
FOOTNOTES 278REFERENCES 299LIST OF MDRC STUDIES IN THEWORK/WELFARE DEMONSTRATION 300
33
LIST OF TABLES
TABLE PAGE1 SWIM AFDC SAMPLE: SHORT-TERM IMPACTS ON EMPLOYMENT,
EARNINGS, WELFARE RECEIPT, AND WELFARE PAYMENTS xxiv
2 SWIM AFDC-U SAMPLE: SHORT-TERM IMPACTS ON EMPLOYMENT,EARNINGS, WELFARE RECEIPT, AND WELFARE PAYMENTS xxv
1.1 DESIGN FOR THE EVALUATION OF SWIM 18
2.1 SUMMARY OF KEY PROGRAM ACTIVITIES 25
2.2 PRIMARY RESEARCH SAMPLES FOR THE PROCESS ANDIMPACT STUDIES 30
2.3 SELECTED CHARACTERISTICS OF REGISTRANTS AT THE TIMEOF INITIAL REGISTRATION, BY ASSISTANCE CATEGORY 32
2.4 SELECTED CHARACTERISTICS OF REGISTRANTS AT THETIME OF INITIAL REGISTRATION, BY ASSISTANCECATEGORY AND WELFARE STATUS 39
2.5 LENGTH OF AVAILABLE FOLLOW-UP BY DATA SOURCE ANDPERIOD OF INITIAL REGISTRATION 47
3.1 PERCENTAGE DISTRIBUTION OF ALL NON-SUPERVISORYAND NON-ADMINISTRATIVE STAFF TIME SPENT ON SWIMTASKS, BY TASK, AGENCY, AND TYPE OF STAFF 54
4.1 SUMMARY OF ALLOWANCES AND SUPPORT SERVICES,BY TYPE OF SERVICE AND COMPONENT 93
5.1 SUMMARY OF PARTICIPATION MEASURES USED IN THESWIM REPORT 98
5.2 TWELVE-MONTH ACTINITY MEASURES FOR SWIM-ELIGIBLES,BY ASSISTANCE CATEGORY 102
5.3 TWELVE -MONTH ACTIVITY MEASURES FOR SWIM-ELIGIBLES,
BY ASSISTANCE CATEGORY AND WELFARE STATUS 109
5.4 PERCENTAGE DISTRIBUTION OF SWIM-ELIGIBLES BY
ASSISTANCE CATEGORY AND PARTICIPATION IN A FIRSTOR SECOND ACTIVITY WITHIN TWELVE MONTHS FOLLOWINGINITIAL REGISTRATION 113
5.5 PERCENTAGE DISTRIBUTION OF CHILDCARE ARRANGEMENTSFOR AFDC PARTICIPANTS, BY PROVIDER AND LOCATION 121
-XXX-
34
5.6 PERCENTAGE DISTRIBUTION OF CHILDCARE ARRANGEMENTSOF AFDC PARTICIPANTS' YOUNGEST CHILD, BY PROVIDERAND LOCATION, AND BY REVIEW MONTH 124
5.7 PERCENTAGE DISTRIBUTION OF CHILDCARE ARRANGEMENTSOF AFDC PARTICIPANTS, BY PROVIDER OF CHILDCAREAND CHILD'S AGE 125
6.1 PERCENTAGE DISTRIBUTION OF SWIM- ELIGIBLES, BY PARTI-CIPATION STATUS AND NUMBER OF MONTHS REGISTERED FORSWIM DURING THE TWELVE MONTHS FOLLOWING INITIALREGISTRATION 129
6.2 PERCENTAGE DISTRIBUTION OF SWIM- ELIGIBLES, BYPERCENT OF MONTHS ACTIVE OUT OF MONTHS REGISTERED,AND BY ASSISTANCE CATEGORY 131
6.3 KEY NONCOMPLIANCE INDICATORS FOR SWIM-ELIGIBLES,BY ASSISTANCE CATEGORY 135
6.4 PERCENT OF NONCOMPLIANT SWIM-ELIGIBLES, BY TYPEOF NONCOMPLIANCE AND ASSISTANCE CATEGORY 1.36
7.1 NUMBER OF REGISTRANTS ELIGIBLE FOR SWIM SERVICES,BY MONTH AND ASSISTANCE CATEGORY 150
7.2 PERCENTAGE DISTRIBUTION OF INDIVIDUALS REGISTEREDWITH SWIM AS OF EACH MONTH, BY NUMBER OF MONTHSREGISTERED WITH THE PROGRAM PRIOR TO EACH MONTH 154
7.3 SELECTED CHARACTERISTICS OF SWIM-ELIGIBLES ATTHE TIME OF INITIAL SWIM REGISTRATION, BY PARTI-CIPATION STATUS IN JULY AND NOVEMBER, 1986 175
7.4 PERCENTAGE DISTRIBUTION OF REGISTRANTS' REASONSFOR NON - PARTICIPATION, BY EXPECTED PARTICIPATIONSTATUS 180
7.5 PERCENTAGE DISTRIBUTION OF SWIM-ELIGIBLES, BYPARTCIPATION STATUS AND RELATED ACTIVITIES ORSTATUSES IN JULY AND NOVEMBER, 1986 185
7.6 ALTERNATIVE ACTIVITY MEASURES FOR JULY ANDNOVEMBER, 1986 187
8.1 ALL AFDC: SHORT-TERM IMPACTS ON EMPLOYMENT,
EARNINGS, WELFARE RECEIPT, AND WELFARE 2AYMENTS 194
-x
8.2 ALL AFDC: SHORT-TERM IMPACTS ON DISTRIBUTION OFEARNINGS, EARNINGS/WELFARE MIX, AND MEASUREDINCOME IN QUARTER FOUR 200
8.3 AFDC APPLICANTS AND RECIPIENTS: SHORT-TERM IMPACTSON EMPLOYMENT, EARNINGS, WELFARE RECEIPT, ANDWELFARE PAYMENTS 204
8.4 AFDC EARLIER COHORT: LONGER TERM IMPACTS ONEMPLOYMENT, EARNINGS, WELFARE RECEIPT, AND WELFAREPAYMENTS 208
8.5 ALL AFDC-U: SHORT-TERM IMPACTS ON EMPLOYMENT,EARNINGS, WELFARE RECEIPT, AND WELFARE PAYMENTS 211
8.6 ALL AFDC-U: SHORT-TERM IMPACTS CN DISTRIBUTIONOF EARNINGS, EARNINGS/WELFARE MIX, AND MEASUREDINCOME IN QUARTER FOUR 215
8.7 AFDC-U APPLICANTS AND RECIPIENTS: SHORT-TERMIMPACTS ON EMPLOYMENT, EARNINGS, WELFARE RECEIPT,AND WELFARE PAYMENTS 216
8.8 AFDC -U EARLIER COHORT: LONGER-TERM IMPACTS ONEMPLOYMENT, EARNINGS, WELFARE RECEIPT, ANDWELFARE PAYMENTS 21)
A.1 SUMMARY STATISTICS FROM QUALITY CHECK OF DATAFOUND IN THE SWIM AUTOMATED TRACKING SYSTEM ........... 233
B.1 SELECTED CHARACTERISTICS OF REGISTRANTS AT THETIME OF INITIAL REGISTRATION, BY ASSISTANCE CATEGORYAND RESEARCH GROUP 242
B.2 SELECTED CHARACTERISTICS OF REGISTRANTS AT THETIME OF INITIAL REGISTRATION, BY ASSISTANCE CATEGORYAND PERIOD OF INITIAL REGISTRATION 244
B.3 SELECTED CHARACTERISTICS OF SWIM-ELIGIBLES AT THETIME OF INITIAL REGISTRATION, BY ASSISTANCE CATEGORYAND YEAR OF INITIAL REGISTRATION 246
C.1 TWELVE-. AND EIGHTEENMO NTH ACTIVITY MEASURES FORSWIM-ELIGIBLES, BY ASSISTANCE CATEGORY 250
D.1 PERCENTAGE DISTRIBUTION OF SWIM-ELIGIBLES, BY NUMBEROF MONTHS ACTIVE OUT OF MONTHS REGISTERED DURING THETWELVE MONTHS FOLLCWING INITIAL REGISTRATION 252
36
D.2 PERCENT OF NONCOMPLIANT SWIM-ELIGIBLES RECEIVINGFOLLOW -UP CONTACT, BY METHODO OF FOLLOW-UP ANDASSISTANCE CATEGORY
D.3 PERCENTAGE DISTRIBUTION OF AFDC SWIM-ELIGIBLES, BYWELFARE, PROGRAM, EMPLOYMENT, AND PARTICIPATIONSTATUS BY THE TWELFTH MONTS FOLLCWING INITIALREGISTRATION
D.4 PERCENTAGE DISTRIBUTION OF AFDC -U SWIM-ELIGIBLES, BYWELFARE, PROGRAM, EMPLOYMENT, AND PARTICIPATIONSTATUS BY THE TW ELFTH MONTH FOLLCWING INITIALREGISTRATION
254
255
257
E.1 PERCENT OF INDIVIDUALS ELIGIBLE FOR SWIM IN JULYOR NOVEMBER, 1986, WHO PARTICIPATED IN PROGRAMACTIVITIES, BY TYPE OF ACTIVITY AND ASSISTANCECATEGORY 260
F.1 AFDC: ESTIMATED REGRESSION COEFFICIENTS FOR EMPLOY-MENT AND WELFARE MEASURES IN QUARTER FOUR 264
F.2 AFDC: EARNINGS AMONG EMPLOYED EXPERIMENTALS ANDCONTROLS 267
F.3 AFDC: SHORT-TERM IMPACTS ON MEASURED INCOME 268
F.4 AFDC EARLIER AND LATER COHORTS: SHORT-TERM IMPACTSON EMPLOYMENT, EARNINGS, WELFARE RECEIPT, ANDWELFARE PAYMENTS 269
F.5 AFDC-U: ESTIMATED REGRESSION COEFFICIENTS FOREMPLOYMENT AND WELFARE MEASURES IN QUARTER FOUR 271
F.6 AFDC-U: SHORT-TERM IMPACTS ON MEASURED INCOME 274
F.7 AFDC-U EARLIER AND LATER COHORTS: SHORT-TERM IMPACTSON EMPLOYMENT, EARNINGS, WELFARE RECEIPT, ANDWELFARE PAYMENTS
-xxxiii-
275
LIST OF FIGURES
FIGURE PAGE
1 PERCENT OF INDIVIDUALS ELIGIBLE FOR SWIM DURINGEACH MONTH WHO PARTICIPATED IN PROGRAM-ARRANGEDACTIVITIES, SELF-INITIATIVED ACTIVITIES OR EMPLOY-MENT, BY TYPE OF ACTIVITY xix
2.1 FLOW OF REGISTRANTS THROUGH SWIM 27
3.1 ORGANIZATIONS, STAFF, AND SERVICES IN THE SWIMDELIVERY SYSTEM, 51
5.1 ILLUSTRATION OF TIME PERIODS COVERED BY TWELVE,-MCNTH LONGITUDINAL ACTIVITY MEASURES USING TENHYPOTHETICAL SAMPLE MEMBERS 100
6.1 FLCW OF SWIM-ELIGIBLES THROUGH THE NONCOMPLIANCEAND AJUDICATION PROCESS 134
7.1 ILLUSTRATION OF TIME PERIODS COVERED BY MONTHLYPARTICIPATION RATES USING TEN HYPOTHETICAL SAMPLEMEMBERS 145
7.2 PERCENT OF INDIVIDUALS ELIGIBLE FOR SWIM DURINGEACH MONTH WHO PARTICIPATED IN PROGRAM - ARRANGEDACTIVITIES, SELF-INITIATED ACTIVITIES OR EMPLOY-MENT, BY TYPE OF ACTIVITY 156
7.3 PERCENT OF INDIVIDUALS ELIGIBLE FOR SWIM DURINGEACH MONTH WHO PARTICIPATED IN PROGRAMARRANGEDACTIVITIES, BY TYPE OF ACTIVITY 159
7.4 NUMBER OF INDIVIDUALS ELIGIBLE FOR SWIM ANDNUMBER OF INDIVIDUALS PARTICIPATING IN SWIM,DURING EACH MONTH 163
7.5 PERCENT OF INDIVIDUALS ELIGIBLE FOR SWIM DURINGEACH MONTH WHO PARTICIPATED IN PROGRAM - ARRANGED
ACTIVITIES, SELF-INITIATED ACTIVITIES OR EMPLOY-MENT, BY ASSISTANCE CATEGORY AND TYPE OF ACTIVITY 164
7.6 ILLUSTnATION OF TIME PERIOD COVERED BY A MONTHLYPARTICIPATION RATE FOR NOVEMBER 1986 USING TENHYPOTHETICAL SAMPLE MEMBERS 166
7.7 PERCENT OF INDIVIDUALS ELIGIBLE FOR SWIM DURINGEACH MONTH WHO PART1"ATED IN ANY ACTIVITY, USINGDIFFERENT ELIGIBILITi CRITERIA 168
38
In
7.8 PERCENT OF INDIVIDUALS ELIGIBLE FOR SWIM DURINGEACH QUARTER OR MONTH WHO PARTICIPATED IN ANYACTIVITY 172
8.1 AFDC: SHORT-TERM IMPACTS ON EMPLOYMENT ANDEARNINGS 196
8.2 AFDC: SHORT-TERM IMPACTS ON AFDC RECEIPTAND PAYMENTS 199
8.3 AFDC-U: SHORT-TERM IMPACTS ON EMPLOYMENT AND
212EARNINGS
8.4 AFDC -U: SHORT-TERM IMPACTS ON AFDC RECEIPTAND PAYMENTS 214
A.1 SWIM CLIENT TRACKING DOCUMENT 226
3D
DEFINITIONS OF ACRONYMS
ABE Ad-lt Basic EducationAFDC Aid to Families with Dependent Children -- zanily GroupAFDC-U Aid to Families with Dependent Children -- Unemployed
ParentCRU Coordination and Referral UnitCWEP Community Work Experience ProgramDEFRA Deficit Reduction Act of 1984ASS Department of Social ServicesEDD Employment Development DepartmentEOPD Employment Opportunity Pilot ProjectEPP Employment Preparation ProgramESL English-As-A-Second LanguageESP Employment Servicet ProgramET Employment and Training Choices Program in MassachusettsEWEP Employment Work Experience ProgramGAIN California's Greater Avenues for Independence ProgramGED General Educational DevelopmentIM Inoame MaintenanceISESA Individualized Supervised Employment Search ActivityJDC Job Developer CounselorsJDU Job Development Unit
JSAP Job Search Assistance ProjectMDRC Manpower Demonstration Research CorporationOBRA Omnibus Budget Reconciliation Ac*'VT On-The-Job TrainingaEACH Realizing Economic Achievement Program in New JerseySSA Social Security AdministrationSTAR Skills Techniques Achievement ReviewsSWIM Saturation Work Initiative ModelWIN 'Tork Incentive ProgramYMCA Young Men's Christian Association
4 0
INTERIM REPORTON THE
SATURATION WORK INITIATIVE MODELIN SAN DIEGO
CHAPTER 1
INTRODUCTION
From July 1985 through June 1987, the County of San Diego in
California operated the Saturation Work Initiative Model (SWIM) as part of
the Social Security Administration's (SSA) two-site Demonstration of
Saturation Work Programs in an Urban Area.1 The other SSA demonstration
site was Philadelphia, Pennsylvania.
The SSA demonstratiOn, directed towards individuals applying for or
receiving benefits under the Aid to Families with Dependent Children (AFDC)
program, was intended to test the feasibility of having at least three-
quarters of program-eligibles active in a welfare employment program at all
times. In addition, the demonstration was intended to measure the
effectiveness of such a program in terms of program-eligibles' future
employment and welfare receipt.
The program model developed by the an Diego County Welfare Department
and the State Department of Social Services included a variety of
activities: job search programs, which taught individuals how to find and
obtain unsubsidized jobs; the Employment Work Experience Program (EWEP), in
which welfare recipients were required to work in public or nonprofit
agencies in exchange for their welfare benefits; and referrals to community
education and training programs.
The program operated in two of the seven welfare employment offices in
San Diego County, constituting about 40 percent of the county's caseload.
Program eligibles included both single-parent AFDC family heads (primarily
-1-
42
mothers) and heads of two-parent families (primarily fathers) in the
Unemployed Parent (AFDC-U) category. 2 Participation in SWIM activities was
required, on an ongoing basis, of all WIN-mandatory AFDC and AFDC-U appli-
cants and recipients. The head of an AFDC-U household is automatically
WIN- mandatory; most AFDC heads of household whose youngest child is at
least six years old are considered WIN-mandatory.3
The Manpower Demonstration Research Corporation (MDRC) has leen
evaluating SWIM under a contract from the California Department of Social
Services. This report, the first of two on the SKIM demonstration,
examines the feasibility of continuously serving a large proportion of WIN-
mandatory individuals in a mandatory, fixed-sequence, multi-component
program. Different ways of measuring participation, the factors that
affected participation levels, and preliminary program impacts are also
presented. A final report, scheduled for completion in 1989, will examine
longer-term impacts and compare the program's benefits with its costs.
In comparison to previous welfare employment programs evaluated by
MDRC, several features of SWIM are especially distinctive. First, the SSA
demonstration was an attempt to set realistic benchmarks concerning the
definition of 'most people participating.' This objective grew out of
increasing interest in making welfare receipt more conditional on
participation in employment- enhancing activities than had typically been
the case. SWIM was intended not only to test whether the arbitrarily set
benchmark of 75 percent was achievable, but to define the feasible upper
bounds of a participation (or saturations) standard. To facilitate this
feasibility test, SSA provided demonstration funding to augment the
county's regular WIN monies and state Employment Preparation Program (EPP)
-2- 43
monies. Although previous welfare employment programs encompassed the
entire WIN-mandatory caseload in specific areas, rarely did they have
either a clear saturation objective or funding above regular WIN funding
-levels.
Second, one of the objectives of SWIM was to require continuous
participation of registrants in program activities for as long as these
individuals remained on welfare. Most other welfare initiatives have had
program requirements of a limited duration or, if the program model
specified ongoing participation, requirements that were de facto short
term.
Third, similar to sane recent state welfare policy initiatives, SWIM
referred individuals to more intensive services than those generally
offered by welfare employment programs. Specifically, SWIM sought to
encourage the participation of welfare recipients in education and training
programs through referrals (not involving any additional funding) to public
and nonprofit community organizations and schools. Th 5 ,MIM evaluation
examines procedures used to set up such lInkages and the participation
patterns in such linked programs.
The rest of this chapter sets the context for understanding the
results of the SWIM evaluation and highlights the background factors that
suggest caution in generalizing the results. The first and second sections
briefly discuss attempts to encourage participation in welfare employment
programs on the national level, as well as welfare initiatives in the State
of California and San Diego County. The third section reviews the develop-
ment and distinctive features of the SWIM program model. The fourth
section describes the program setting. The final section summarizes the
-3- 44
salient features of MDRC's evaluation design
I. Participation in Previous Welfare Employment Programs
The federal Work Incentive (WIN) program, created in 1967, was
intended to provide skills assessment, job training, placement and support
services to help AFDC recipients became self-supporting. Originally
introduced as a voluntary program, WIN became mandatory in 1971; that is,
in order to receive AFDC benefits, all adult recipients without preschool
children or specific problems that kept them at have had to register with
the state employment service, participate in available job training or job
search activities, and accept employment offers.
Despite these provisions, a relatively small share of mandatory
registrants received employment and training services. The program's
inability to enforce its participation requirement is generally attributed
to two factors -- inadequate funding to operate sufficient activities for
the caseload of mandatory registrants and the discretion allowed program
operators to grant exemptions and deferrals. Federal restrictions also
limited state and local agencies' ability to modify WIN's program
regulations.
Since WIN's establishment, many state and local agencies have experi-
mented with different employment and training approaches for welfare
recipients. These typically have required additional funding and/or
waive' of WIN regulations fran the federal government. In the 1970s,
thes, programs reached varying proportions of the AFDC caseload, but
gener'Ally had modest rates of participation.4
Passage of the Omnibus Budget Reconciliation Act (OBRA) in 1981 marked
-4-
45
an important milestone in the development of welfare employment policy.
OBRA and related legislation permitted the states -- as part of their
regular WIN programs -- to require applicants for AFDC and AFDC-U to
participate in job search assistance and recipients to take Community Work
Experience Program (CWEP) assignments as a condition of receiving welfare
benefits. (Recipients had been subject to job search and other require-
ments prior to OBRA.) In addition, the WIN Demonstration provisions
increased states' flexibility in designing and managing their WIN programs.
This included administration by a single agency instead of the previous
dual agency structure, under which state employment agencies were
responsible for training and employment and :.Late welfare departments
provided support services.
More than half the states have responded to the OBRA flexibility and
established programs -- usually in selected areas rather than statewide --
that require welfare recipients to participate in job search and/or work
experience activities.5 MDRC's Demonstration of State Work/Welfare
Initiatives examined the effectiveness of post-OBRA programs in eight
states. In most of the states studied, participation rates were above
those achieved in previous special demonstrations or in the WIN program.6
Typically, within six to nine months of registering with the new program,
about half of the AFDC group had taken part in some activity for at least
one day, and substantial additional numbers had left the welfare rolls and
the program. The programs generally led to modest increases in employment
that in sane cases were associated with welfare savings. The impacts were
usually large enough to offset the programs' costs, though not for every
target group in every state.
-S-
4U
Most of the programs studied, unlike SWIM, had participation require-
ments that were of a short-term nature, in practice if not by design. By
far the major activity was job search, a relatively short (usually no more
than two to three weeks) and inexpensive intervention. Education and
training activities were limited. And work experience, when required, was
almost always a short-term obligation, usually lasting no more than 13
weeks.
One exception to this pattern was West Virginia. In 1982, this state
established a statewide unpaid work experience program (which is still
operating, primarily as a work experience program) with an ongoing
participation requirement -- a straightforward work program, in which the
assignment lasts as long as the recipient receives welfare. The state
successfully imposed the requirement for the heads of two-parent (AFDC-U)
households, but did not impose it rigorously for single parents. In a
demonstration effort designed to saturate the AFDC-U caseload, the program
achieved participation rates of between 59 and 69 percent of the AFDC-U
caseload on a monthly basis.? Recently, other states have begun implement-
ing programs designed to emphasize more intensive services or requirements
(including education and training), and to complement these with extensive
childcare services.8 SWIM represents one program variation of introducing
comprehensive services and/or longe.: participation obligations.
II. Welfare Initiatives in the State of California and San Diego County
California was involved in several welfare employment initiatives
prior to the 1980s. Between 1972 and 1975, California operated a work
experience demonstration for AFDC recipients as part of the California
47-6-
Welfare Reform Act. Under this demonstration, the state's Employment
Development Department (EDD) was directed to create work positions in
public agencies for all mandatory WIN registrants who were not assigned to
other activities. However, implementation was hampered by limited funding
and legal challenges; during 1974 the program placed less than 3 percent of
eligible registrants in program work positions.
After the 1974 elections and a subsequent change in administration,
the California legislature repealed the state's authority to test community
work experience for the welfare population and substituted a new set of
employment and training services which focused on job clubs. Consequently,
EDD and the Department of Social Services (DSS) developed the Job Search
Assistance Project (JSAP) -- a demonstration implemented in 197 5 to offer
welfare applicants both group and individual job search training, and same
skills training.
Tcward the end of 1979, JSAP was expanded in the form of a bill
seeking to distinguish between semployaFle and 'nonemployable' welfare
recipients and calling for early intervention to prevent employable persons
from becoming long-term recipients. The primary service was to be group
job s -'arch, as used in JSAP, with the promise of training for those who did
not find work through job search.9 In 1980, the California legislature
authorized the Employment Preparation Program (EPP) which was initially
implemented on a demonstration basis in three counties: Lake, San Mateo
and Ventura. After the 1981 passage of OBRA, the California legislature
continued to emphasize job search assistance approaches and to reject
statewide work experience proposals.
A program that did involve work experience was developed in San Diego,
-7-
however. This was a test of EPP job search followed by community work
experience (called the Experimental Work Experience Program, or EWEP) for
those who failed to find jobs through group job search workshops.1° San
Diego County's interest in work programs was based on several factors.
First, the AFDC-U and, particularly, the AFDC caseloads had grown steadily
over the last decade. Second, the county had already experimented with
workfare programs for recipients of other income tiansfer programs:
General Relief and Food Stamps. Third, the county perceived that there was
strong public support for a work-for-benefits approach and also considered
itself a leader on issues of welfare reform.
In designing the program, San Diego County officials specified two
main objectives: developing the work skills of welfare recipients, and
reducing the rolls and costs of welfare. In pursuit of these objectives,
the program was structured as a sequential program with four stages. The
first stage was job placement assistance, provided on the day of welfare
application. Following this, people were referred to a three-week group
job search program, where they went through the second =Id third stages.
During the first week, they participated in workshops designed to build
self-confidence and job-seeking skills. In the following two weeks,
applicants were involeed in self-directed job search, using telephone banks
to call prospective employers. Individuals who had not found employment by
the end of the workshop were then referred to EWEP; this was the fourth
stage, in which they were required to hold positions in public or nonprofit
agencies for up to 13 weeks.11
The San Diego initiative began by giving priority to new WIN-mandatory
applicants for the AFDC-U program (primarily males). The target populatio
49-8--
was later expanded to include applicants for AFDC (primarily females).
fund the project, the county became part of the state's existing
three - county EPP demonstration of job search and obtained separate
legislative authority to operate a community work experience program
through a federal demonstration project. Administrative and operational
responsibility for the EPP job search program was assumed by EDD staff,
while county DSS staff administered and operated EWEP. With a clear
mandate to curb welfare caseloads and costs by improving the unsubsidized
employment of applicants to welfare, the project began operations in August
1982. The EPP job search workshops began immediately; FWEP operations
started up in November 1982.
MDRC evaluated the effectiveness of two program sequences: job search
alone (the first three stages) and job search followed by EWEP. The
results indicated that the program successfully implemented its short-term
participation requirement. Approximately 55 percent of AFDC and 60 percent
of AFDC-U registrants participated in sane :ctivity within nine mo.ths of
application.12 As would be expected in a sequential program, among
registrants eligible for both job search and work experience, more garti-
cipated in the farmer than the lacter tabout 55 percent versus 17 percent).
As intanded, EPP/EWFP staff rigorously enixced a mandatory
participation requirement. Program staff succeeded in working with all but
a small proportion of progra -eligible individuals. By nine months after
application for welfare, over 90 percent of the research sample had either
fulfilled program requirements, found jobs, been deregistered from the
program (because they were no longer WIN-mandatory or had been sanctioi
for not cooperating with the program) or had left the welfare rolls.
-9- 5 0
Among A'iC applicants, the job search/EWEP sequence, although not job
search alone, led to increases in employment and earnings and modest
welfare savings. Among AFDC-U applicants, there were no statistically
significant impacts on employment and earnings, but substantial reductions
in welfare payments under both program sequences.
From the perspective of government budgets, operating costs were
offset by Lnefits (in terms of reduced welfare and MediCal payments,
increased taxes and other budget gains) for AFDC and AFDC-U registrants in
both program sequences. Fran the perspective of the welfare applicants,
the results were not as consistent. For the AFDC applicants assigned to
job search and EWEP, there were clear financial gains; for the AFDC-U
applicants in both program sequences, there were overall losses.13
Encouraged by preliminary MDRC research findings similar to those
described above, the county continued to operate the EPP/EWEP program for
applicants in all areas of San Diego until 1985. EPP continued to operate
in selected counties (6 as of 1985) throughout the rest of the state as
well.
During 1985, several welfare policy changes occurred in California.
First, the state became part of the national WIN Demonstration Program.
This transition, which occurred in July 1985, changed the institutional
arrangements for delivering employment and training services, and allowed
greater flexibility in how these services were combined.14
Second, the Greater Avenues for Independence (GAIN) program, a major
new welfare employment initiative, was passed by the legislature. The GAIN
legislation was developed over a period of five months by a coalition of
liberals and conservatives and passed by the state legislature in late
1985. Counties were given up to three years to design their GAIN programs.
Also in 1985, independent of the above two developments, San Diego's
EPP/EWEP model changed into SWIM. As noted, SWIM was part of SSA's
Demonstration of Saturation Work Programs it an Urban Area, and operated in
the two most urban welfare administrative areas of San Diego county. The
other five welfare administrative areas in the county continued to operate
EPP/EWEP until 1987, at which time San Diego implemented the GAIN program
countywide. 15
The SWIM program began operations in July 1985, after a year of
planning.16 In response to SSA's June 1984 announcement of a saturation
demonstration grant, the California Department of Social Services and the
County of San Diego Department of Social Services submitted a proposal for
SWIM, noting that of the fifteen qualifying urban counties in the state,
San Diego was the only county that had the requisite experienle in
operating employment programs to be able to undertake a saturation work
program a large scale within a reasonable timeframe. In September 1984,
SSA chose San Diego County as one of the saturation demonstration sites.
According to San Diego County officials, the county had several
objectives in applying for the saturation demonstration grant. First, the
county viewed the demonstration grant as an opportunity to obtain general
funding for their welfare employment programs in the face of declining WIN
monies and an anticipated decline in EWEP funding.
Second, the grant would allow the county to emphasize education and
training to a greater extent than was possible in EPP/EWEP. The county
viewed the demonstration grant as an opportunity to obtain funding for the
number of sta'f deemed necessary to add this type of component to the
52
EPP/EWEP model.
Third, early results from the EPP/EWEP evaluation had indixated that
the program was effective in increasing the employment levels of WIN-
mandatory applicants and decreasing welfare costs. County officials viewed
the demonstration grant as a means of funding an evaluation to determine
the relative effectiveness of requiring participation of the entire WIN-
mandatory caseload as opposed to only applicants.
III. The SWIM Program Model and Funding Resources
The SWIM program model built upon the county's previous experience
with welfare employment programs in several ways. First, as noted above,
SWIM extended the EPP/EWEP model to recipients as well as applicants, thus
targeting the program on the entire WIN-mandatory caseload. In their grant
proposal to SSA, the county noted the existence of a pool of more than
13,000 medium- and long-term AFDC recipients who had received very limited
services through the current programs.
Second, SWIM replaced the EPP/EWEP short -term participation require-
ment with a continuous participation requirement that was to last as long
as an Individual remained registered with WIN. To accomplish this, the
county added several components to the EPP/EWEP sequence. These components
-- which were available to registrants who completed job search work '`tops
and EWEP without finding a job -- included Adult Basic Education, General
Educational Development (GED) test preparation, English as a Second
Language (ESL) programs, skills training, on-the -job training and addition-
al job search activities. It is important to note that the program itself
did not operate or fund education or training activities. Rather, staff
53-12-
referred program registrants to already existing community programs.
Third, SWIM added a new set of staff, knc,'n as the Coordination and
Referral Unit (CRU), to the EPP/EWEP staff configuration. The CRU Staff
had primary responsibility for monitoring participants' progress and
continuously assessing participants' needs for 'employment service
intervention. Assessments occurred most commonly once a registrant had
completed the job search workshop/EWEP sequence.
Fourth, biweekly job clubs, which often operated concurrently with
EWEP, were established. Two-hour job search workshops, held once a week
for a period of 13 weeks, were also added to the model. Registrants could
:e referred to this latter component, known as ISESA (Individu lized
Supervis 1 Employment Search Activity), __ter completion of the job club
component and/or EWEP.
Several other program changes were made as conditions of a state
legislative waiver which allowed the county to continue to operate EWEP
from July 1935 through June 1987. Sanctioning rules for AFDC-U registrants
who were noncompliant in EWEP were made less punitive. Only the head of
the case lost AFDC benefits when a sanction was in connection with EWEP
requirements.17 The EliEP work hors obligation, which had previously been
computed by dividing the registrant's AFDC grant by the federal minimum
wage, was changed to use prevailing wage rates rather than federal minimum
wage rates. 18 Finally, additional conciliation -- counseling of
registrants and 'secoud: chances wcis required prior to the application
of an EWEP sanction.
GAIN and SWIM are similar in that both programs have multicomponent
models and involve a continuous participation requirement. Howev_tr, GAIN
5-13-
differs from SWIM and from other past welfare initiatives in several
important respects. These differences should be kept in mind when
interpreting the SWIM findings.
First, and probably most important, GAIN mandates remedial education,
early in the model, for those who fail a diagnostic test or lack a high
school diploma or GED. Second, GAIN moves away from one prescribed
sequence of program activities to the prescription of various sequences
determined by registrant characteristics. Third, GAIN uses a registrant
contract to provide some registrant choice regarding services, to ensure
the provision of program services, and to emphasize the registrant's
obligation to participate. Fourth, GAIN provide;: payments to community
educatioh anA training agencies who serve GAIN registrants. Lastly,
although SWIM could provide some support monies to registrants, GAIN can
provide substantially more. In particular, childcare monies are available
to individuals participating in self-initiated activities and, for a short
time, to those who find jobs while in the program.
GAIN is intended to operate on a much wider scale than SWIM. Because
SWIM operated in only two welfare administrative areas instead of county-
wide (and placed education and training at a later point in the program
model), it did not test the capability or capacity of community organi-
zations to absorb large nuabers of welfare recipients i-to their programs.
It is important to note that as a federal demonstration site, San
Diego received special demonstration fund-41g -- 95 percent paid by SSA and
5 percent paid by the State Department of Social Services. The non-
research portion of these funds, which was provided in addition both to
regular WIN and to special EPP funding in the county, amounted to apt.roxi-
-14-
mately $1,700,000 over a three-year period.19 In addition, the program
depended on community resources to provide education and training services
and did not fund these.
These resources covered the approximately 9,200 individuals who
registered with the program and were eligible for program services (i.e.
they were not assigned to the control group) in the 27 months that SWIM
operated.
IV. Program Setting
As explained in the previous section, San Diego was an unusual setting
in which to test tte feasibility of operating a multi-component saturation
program with an ongoing participation requirement. Its welfare department
had extensive exverience operating welfare employment programs before
initiating SWIM. This experience included coordination among numerous
agencies outside the welfare department ane staffing units within the
department. Several other aspects of the program setting were also
important.
With a 1986 population of 2,166,200, San Diego County is the second
most populous county in the State of California." The City of San Diego
(population 1,002,900 in 1986) is the second largest ci in the state and
the eighth largest city in the country. Located next to the Mexican
hder, the county has a high proportion -- 18 percent -- of non-English
speaking residents. Less than 7 percent of the residents live in rural
areas. In comparison to other California counties, a high proportion -- 78
percent -- of the population are high school graduates. Eleven percent of
the population lives in poverty.
-15-
The county has an extensive network of educational facilities. In
addition to one university and one college of the state's public higher
education system, the county has five community college districts. Most
SWIM registrants lived within the jurisdiction of the largest of these, the
San Diego Community College district. The county also has eight different
adult school districts. Unlike the usual situation in the county, most of
the adult schools in the SWIM areas are under the purview of the local
community college district, i.e. the San Diego Community College district.
The availability of extensive education -nd training opportunities in
the county increased the likelihood that registrants could, on their own,
enroll in these community programs. In fact, according to information
gathered at initial program registraJon, approximately 15 percent of the
SWIM AFDC registrants and 10 percept of the AFDC-U registrants were in
these types of programs as of program entry. 21 The existence of this
network of education and training programs also facilitated the placement
of registrants in these activities by the SWIM program.
During the period when SWIM was operating, the local economy was
relatively healthy. Unemployment rates in the county were 6.5 percent in
1984; 5.3 percent in 1985; 5.0 percent in 1986; and 4.3 percent in May 1987
-- all below the prevailing rates for the State of California and the
country as a whole.22 The county's economic base contains a wide variety
of industries, including services (25 percent), wholesale and retail trade
(23 percent), government (18 percent) and manufacturing (15 percent).
AFDC grant levels in California are relatively generous, ranking tenth
highest in the nation in 1986. At the start of the SWIM program, a family
of three with no other income was eligible for $587 per month. This was
increased to $617 in July 1986 and $633 in July l!':87.
Welfare recipients can combine work and welfare if they meet
eligibility tests for AFDC and have earnings that do not exceed a state's
payment standard or grant level after allowable deductions. 23 The healthy
economy and high AFDC grant levels in San Diego enabled many registrants to
combine unsubsidized employment with receipt of welfare.
SWIM operated in the county's two most urban EPP administrative are&s,
representing a population base of approximately 487,000 individuals,
containing approximately 40 percent of the county's welfare caseload and
encompassing subareas with heavy concentrations of low-income individuals.
Data on the characteristics of applicants in the two offices that operates'
SWIM compared with those of applicants in the other five county offices are
available for 1982-1983 from MDRC's evaluation of the EPP/EWEP program.
Applicants in these two offices, compared to the other applicants, were
more likely to be black or Hispanic; were less likely to have a high school
diploma or GED; had slightly longer welfare histories; and were less likely
to have been employed in the year prior to application. These differences
were more evident among the AFDC applicants than the AFDC-U applicants.
V. Evaluation Design: An Overview
MDRC's evaluation of SWIM comprises three parts: process or implement-
ation, impact and benefit-cost analyses. Table 1.1 shows the key
questions, methodology and data sources associated with each. This report
presents process and preliminary impact findings; the final report will
present longer-term impacts and benefit-cost results.
-17- r8
TABLE 1.1
DESIGN FOR THE EVALUATION OF SHIM
Research Component and QuestionsI
Methodology Data Sources
PROCESS ANALYSIS
Within a set period of time from program Analysis of patterns of program assignment. Program administrative records, includingregistration. what patterns of program partici-
potion existed and what factors explain
observed differences? Did participation rates
vary for different subgroups of the popula-
tion?
participation, and deregistrotion status, outcome, and participation data
Was participation mandatory and did individuals
participate on on ongoing basis?Vudy of the interaction between porticipa-
tion patterns and program design. institu-Systematic observation, case file studies,
interviews with program staff, programtutional arrangements, administrative prac-
tices, and other conditionsadministrative records
In any month, what percent of those eligible Analysis of patterns of program registration. Program administrative records, includingfor the program were participating? participation and deregistrotion status, outcome and participation data
What is the content and administrative struc- Study of program components and staff Systematic observation, interviews withture of the demonstration program? decision-making program staff, program administrative
records
IMPACT ANALYSIS
Did SHIM result in an increase In employment Comparison of the employment and welfare Uniform client characteristics collectedand earnings and/or a reduction in welfare outcomes over Jme for AFDC and AFDC-U at sample entrydependency and benefits? applicants and ruiplents randomly assigned
to the experimental treatment or to a centre! AFDC payment files, Unemployment InsuranceDid impacts vary for AFDC and AFDC-U
registrants or far other subgroups?group receiving no program services earnings files
Program administrative records
(continued)
TABLE 1.1 (continut.1)
Research Component and Questions Methodology Data Sources
BENEFIT-COST ANALYSIS
Did SWIM lead to an increase or decrease in
direct budget expenditures?
Did SWIM make the experimental group better
off financially?
Estimation of the increment or decrement of
operating costs (Including administrative
costs and payments to institutions and to
participants for work related expenses) for
experimentols compared to the control group
Estimation of the rat present value of 9N IM
by comparing additional costs and benefits
State and loco! budgets. data on special
payments and studies of staff time ollico-
tion
Cost date. program odministrorve records.
Impact estimates. and value of -itput
estimates from Interviews with work exper-
ience supervisors
A. The Process Analysis
The process analysis examines the operation of SWIM and identifies the
factors that facilitated or constrained implementation. The analysis has
three main parts. The first describes the content and operations of the
program, highlighting its major activities rind administrative procedures.
The second part analyzes the movement of registrants through the program,
examining participation patterns for groups of registrants throughout a
uniform follow-up period -- 12 months after registration.
A third part of the analysis examines participation by providing
'snap-shots' of program operations at set points in time. One type of snap-
shot, for example, can show program operations in a specific month of SWIM
by examining the proportion of those eligible during that month who were
participating during that month. Another type can show operations in a
month by disaggregating participants according to the types of activities
in which they participated during the month. These types of snap-shots for
each of the 24 months of SWIM indicate the degree to which the program
saturated the WIN-mandatory caseload over the course of the demonstration.
B. The Impact Study
Tile impact analysis measures the effects of SWIM on the employment,
earnings and welfare receipt of registrants. To estimate program impacts,
an experimental design was implemented during the first 12 months of the
program. During this pericd, individuals in the existing WIN- mandatory
caseload of the two SWIM offices, along with any individuals determined to
be WIN-mandatory during that year, were randomly assigned to one of two
research groups. Members of the experimental group were required to parti-
cipate in SWIM; members of the control group were not assigned to SWIM
-20-
63
activities but could, on their awn initiative, enroll in community
programs.
Because the evaluation was intended to test the feasibility of
saturating a WIN-mandatory careload, only the minimum number of individuals
required to provide reliable estimates of impacts -- 30 percent of the AFDC
registrants and 35 percent of the AFDC-U registrants -- were assigned to
the control group during the 12-month period. After the 12month period of
impact sample intake, assignment to the control group stopped and all new
registrants were subject to SWIM participation requirements. Since
successful random assignment ensures that experimental and control group
members are similar in all characteristics except eligibility for program
services, any differences in the groups' experiences result from
differences in program treatment. that is, tie requirement to participate
in SWIM services the receil... of these services.
Impacts were estimated by comparing the welfare and employment experi-
ences of all experimentals (regardless of whether they had participated or
were employed while in the program) and controls over time.
In evaluations of programs with limited participation requirements, a
12-month follow-up period will generally include substantial post-program
follow-up data on both employment and welfare receipt. Since SWIM
registrants were supposed to participate for as long as they received
welfare, there could be no post-program follow-up on those who were still
on welfare and active in SWIM throughout the follow -up period. In fact, 15
percent or the AFDC registrants and 14 percent of the AFDC-U registrants
were still active in job search, work experience or education/training at
tkle end of the 12-month follow-up period available for analysis in this
64-21-
report. And almost half the registrants in the impact sample -- 47 percent
of the AFDC's and 42 percent of the AFDC-U's -- were still registered with
SWIM at this point and eligible for SWIM services, even if not currently
active.24 Longer follow-up (11 the impact sample will be analyzed in the
final SWIM report.
C. The Benefit-Cost Analysis
This analysis, which will appear in the final report, will examine the
differences between the benefits and costs for the experimental group and
those for the control group. Operating costs, including program
administration and staff costs, and direct payments to enrollees and to
institutions and organizations, will be compared to net benefits. The
latter include net reductions in welfare grants or other transfer program
payments, as well as net increases in the taxes paid by individuals who
became employed as a result of the program.
The remainder of this report consists of seven chapters. Chapter 2
describes in detail the research design, the characteristics of the
research sample and data sources for the evaluation. Chapter 3 discusses
the program model, administrative structure and staffing patterns in SWIM.
Chapter 4 provides detailed descriptions of the nature of the services
provided in SWIM. Chapter 5 examines participation patterns, from entry
into the program to one year later, for the sample of registrants on which
the impact analysis is based. This provides information on the types,
sequences and duration of services, and includes discussion of the typical
-22-
6 5
childcare arrangements for program participants. Chapter 6 examines the
extent to which the program implemented an ongoing participation
requirement as well as noncompliance activities and outcomes. Chapter 7
analyzes monthly participation rates in the program. Chapter 8 concludes
the report by considering the program's short-term impacts on employment,
earnings, welfare receipt and welfare payments.
-23-eCJ
CHAPTER 2
RESEARCH DESIGN, SAMPLES AND DATA SOURCES
This chapter presents the program model tested in the evaluation. It
then describes the process by which the main research sample was randomly
assigned to experimental and control groups. It goes on to examine the
characteristics of the sample used for the impact analysis, and subgroups
of that sample. It then describes the sample used for the monthly parti-
cipation analysis and smaller samples used in other supplementary studies.
The chapter ends with a brief discussion of the data sources.
I. Program Model
As noted in Chapter 1, the SWIM demonstration explicitly tested the
feasibility and effectiveness of imposing an ongoing work requirement on at
least three quarters of the eligible welfare population. The SNiM program
operated in the two largest offices -- San Diego West and Service Center --
of the seven EPP offices in San Diego. These were the two most urban
offices in the county and served the most disadvantaged caseload. The
program model, as seen in Table 2.1, involved a vari,ty of activities,
including job search workshops; EWEP; job clubs; ISESA (Individualized
Supervised Employment Search Activity); program-arranged education or
training; self-initiated education Jr training; and employment, as long as
work hours were between 15 and 30 hours a week.1 The nature of these
program services is discussed in detail in Chapter 4.
-24-
67.10
ORIENTATION
APPRAISAL
JOB SEARCH WORKSHOP
BE P
JOB CLUB
ST.2
ASSESSMENT
ISESA
TABLE 2.1
9e1k1
SUMMARY OF KEY PROGRAM /!:1 IV !TIES
Occurred before any SWIM activity ond included program registration.Individual exit conferences were conducted for c.mtrol group members
directly fol lading orientotion.
Immediately followed orientation and resulted In rAferrol to programactivity. deferrol tram program activities due to participation inapproved self - Initiated activites or zieferral frcm all program
activities.
A two-week activity, provided to registrants WM- orientotion andappraisal. The first week involved group sessions, followed by o
week of phone roan activities.
The ,inployment Work Experience Program (REP) involved unpaid work ata public or non-prof it agency or organizotion. while registrontscontinued to receive their welfare grant. Registrant:, were scheduledfor o moximun of 32 hours eoch week. for 13 weeM.
Biweekly two-hour sessions, usually opercted concurrently with EWEP.
Skills Techniques Achievement Reviews (STAR) replace'. J ob Clubs es ofJanuary 1987 and involved supervised job scorch with groupmotivational sessions for two to three hours every other Coy.
Conducted by progran stoff, after the ccmpletion of EWEP or JobClubs, In order to refer registrants to further job search. educationor training.
The Individual ized Supervised Employment Search Activity (ISESA).uszat ly offered is a post-osseszent activity, required attendonce ofweekly job search sessions for 90 days.
ECU CATION/TRAINING Education and training could be either self-initiated orprogram- arranged. Self - initiated education or training could occurof any point in the model.' it approved by progran staff, activitiesdeferred registrants frcm other progran requirements.Program-arranged education or training usually occurred afterassessment.
EMR.OYILBiT Urr,ubsidized employment could occur at any point in the program. Ifemployed IS to 30 hours a week, o registrant wos deferred frcm otherprogran requirements. If employed less thon 15 hours per week,registrants were given odditional program ossisrments. FlgistrontsLop! ed more than 30 hours a week were dereplflered.
256
.e,
St
..
/
II. Random Assignment
To isolate the impacts of SWIM from the effects of other factors on
employment and welfare receipt, MDRC implemented a randan assignment
design. Randomly assigning registrants to research groups for the purposes
of comparison should ensure that members of the experimental and control
groups are similar in all characteristics except the services they receive.
Any differences in the behavior of the groups, therefore, should be due to
the program treatment being evaluated.
In SWIM, registrants were randomly assigned to two groups: a SWIM-
eligible group and a control group. The members of the SWIM-eligible group
were required to participate in SWIM. Members of the control group could
not receive SWIM services but could enroll on their own initiative in other
community programs, such as Job Training Partnership Act (JTPA) or
community college services.
Random assignment took place at the point of initial registration for
SWIM (see Figure 2.1). The random assignment period began on July 1, 1985,
and ended on June 30, 1986. During the randan assignment period, three
groups of individuals were required to register for SWIM and, thus,
included in the research. One group included individuals who applied for
welfare and were determined to be WIN-mandatory. These individuals are
referred to as applicants throughout the report, even if their applications
were approved and even if they eventually left welfa_e. All AFDC-U parents
are automatically WIN mandatory; most AFDC heads of household whose
youngest child is at least six years old are considered WIN-mandatory.
WIN-mandatory applicants were required to register for SWIM before they
were approved to receive welfare. If their applizations were denied, they
-26-
69
w1:4-MandatoryApplicants
FIGURE 2 . 1
FLOW OF REGISTRANTS THROUGH SWIM
New TM-Mandatory waft-Mandatory RecipientsRedetermined Recipients
)Renewed For Welfare Eligibilit
SWIll Registration. ClientInformation Sheet Completion
end Orientation forExperimental* and Control*
Random Assignmen
Control,
ExitConference
1 No SKIM Scrsicese
(Experimental, )
Appraisal
InSelf-InitiatedEducation or
Training? "
Previouslyin Job Search
Work shopand SIMI.?
no
Employment
Self-Initiated IEducation or
Training
Biweekly 1
_sod ET' withBiweekly Job Clubsb
ISESA.TralnIng. orEducation
NOTES: a Controls could receive services outside of the SWIM program, e .g. communitycollege or JTPA services.
b h January 1987, job clubs were replaced by the STAR component . Registrantsparticipated in STAR after completing job search workshops and before beginning EWEP. Addtionaly,
registrants participated initialy in STAR if they had previously participated in job search workshop.
27 70
were deregistered fran SWIM. £he other two groups included in the research
were WIN-mandatory welfare recipients, also phased into SWIM during the
random assignment period. One of these recipient groups, called redeter-
mined recipients, consisted of welfare recipients who had just been
determined to be WIN-mandatory, generally because their youngest child had
turned six years old. The other group, called renewed recipients,
consisted of welfare recipients who had previously registered for WIN /EPP
but were renewing their registration. This renewal was required every 12
months after the most recent AFDC approval.
Random assignment proceeded as follows. At SWIM registration, local
program staff completed a one-page interview document (called a Client
Information Sheet) eliciting demographic characteristics from all
registrants. Local office staff then telephoned county staff at a central
DSS office to relay a registrant's identifying information. Central office
staff would then assign each registrant to an experimental or control
status, using a list of randomly generated codes supplied by MDRC.
ensure that all registrants remained in the group to which they had
been randomly assigned, even if they were deregistered fran the program and
later re-registered, local office staff as well as central office staff
maintained an alphabetic registrant master log with the research group
status of everyone who had registered with SWIM and been randomly a. signed.
SWIM - eligibles who moved to an area of the county served by one of the five
non-SWIM offices were eligible for EPP/EWEP services. Controls who moved
to those areas of the county were not eligible for any program services.2
As noted, one of SWIM's primary goals was to test the feasibility of
serving at least 75 percent of its SWIM-eligible caseload in each month.
-28-
To test this 'monthly participation' goal, the SWIM-eligible caseload
needed to be as large as possible. For this reason, the size of the
control group was set as the minimum number of individuals required to
provide reliable estimates of impacts. Thirty -five percent of the 4,626
AFDC's who registered with SWIM during the random assignment period were
assigned to the control group. This yielded a total of 1,619 AFDC
registrant controls. Thirty percent of 2,277 AFDC-U's who registered with
the program during the random assignment period were assigned to the
control group. This yielded a total of 683 AFDC-U registrant controls.
Thus, about one-third of those who registered with SWIM during the random
assignment period were not eligible to receive SWIM services, because they
were assigned to control group status.
The remaining 65 percent of the AFDC and 70 percent of the AFDC-U SWIM
registrants were eligible for all SWIM program services and were assigned
to the SWIMeligible group. MDRC collected data for approximately half of
these SWIM- eligibles, 1,608 AFDC's and 7n4 AFDC-U's. This was a random
sample of all SWIM - eligibles, and SWIM program operators did not Allow which
of the SWIM - eligible registrants were in the sample and which were not.
The impact sample, therefore, consists of these 2,312 experimentals and
2,302 controls, approximately two-thirds of all those who registered with
SWIM during the first year of program operations.3
Several different samples were used in the research. The primary ones
are the impact sample; the 12-month activity measure sample which is the
impact sample minus the control group; and the monthly participation
sample. The particulars of these samples aLe summarized in Table 2.2.
They are discussed in more detail in the following sections.
-29-
TABLE 2.2
SMIM
PRIMARY RESEARCH SAMPLES FOR
THE PROCESS AND IMPACT STUDIES
Sample
Registration
Period
Control Group
Included
Total
Sample Size
Chapters
in Which
Sample
is AnalyzedAFDC AFDC -U
Twelve-Month Activity July 1, 1985- no 1608 704 5,6
Measure Samples June 30, 1986
Monthly Porticbpation July 1, 1985- no 533^c
2949c
7
Measure Sample June 30, 1907
Impact Sample July 1, 1985- yes 3211d
1341d
8
June 30, 1986
NOTES:aThis sample is used to examine the extent to which individuals participated in
various activities during the 12 months following initial registration.
bThis sample is used to examine the proportion of individuals eligible for the
program during a month who were active during that month.
cThese sample sizes are weighted 'o reflect the actual number of SHIM- Eligibles and
the octual proportion AFDC's and AFDC-0's who registered between July 1, 1985 and June 30, 1987.
See Chapter 2, footnote number 4, for weighting factors.
dSixty-two individw,s, whos4 initial registration date occurred during the impact
sample intake period, were excluded from the impoct analysis because they did not have social
security numbers. These registrar.', were included in all other analyses.
30 7'3
III. Characteristics of the Impact and 12-month Activity Measure Samples
Table 2.3 presents the characteristics of the impact sample, as r.:port-
ed at random assigit..ent, higi.lighting the differences between AFDC's and
AFDC -U's.4 A full 60 percent of the AFDC-U's are applicants, compared with
only 39 percent of the AFDC's. AFDC-U's tend to be married males living
with their spouses; AFDC's tend to be females who are not married or
married but not living with their spouses. Over 70 percent of AFDC-U's
have children under six, compared with only 10 percent of AFDC's. This is
'lecause AFDC-U's are in families with two-parents, one of whom is required
to participate in SWIM regardless of the age of the children in the home.
AFDC's, in contra7t, are in single-parent families with at least one child
in the hare; if a child is under 6 years old, the parent is not required to
participate in SWIM. The two assistancc groups also differ on ethnicity.
Forty-two percent of the AFDC-U's are hispanic; 42 percent of the AFDC's
are black. About a quarter of bnth groups are white. The highest school
grade completed, on average, was approximately the tenth grade for both
groups.
Thirty-four percent of the AFDC-U's have never had a welfare case in
their own name, canpared with only 11 percent of the AFDC's. Only 15
percent of the AFDC-U's have had a welfare case in their awn name for five
years or more, canpared with 51 percent of the AFDC's. The average number
of months ever on welfare is only 24 for the AFDC-U's, compared with 70
months for the AFDC's.
At the time of registration, 13 percent of the AFDC's and 9 percent of
the AFDC-U's reported neing employed. Only 28 percent of the AFDC-U's
TABLE 2.3
SWIM
SELECTED CHARACTERISTICS OF REGISTRANTSAT THE TIME Of, INITIAL REGISTRATION, BY ASSISTANCE CATF1ORY
Characteristic AFDC AFDC-U
Office (s)
Service CenterSon Diego West
AFDC Status (%)
49.9
50.1
50.3
49.7
Applicant 39.3 59.8***Renewed Recipient 32.9 23.1***Redetermined Recipient
a27.9 17.1***
Average Age (Years) 34.2 32.8***
Sex ( %)
Male 8. 91.3***Female 91.3 8.7***
Ethnicity (%)
White, Non-Hispanic 27.2 24.7*Black, Non-Hispanic 42.2 20.1***Hispanic 25.7 42.1***American Indion/Alcskon Native 0.6 0.4
Asian and Pacific Islander 3.8 11.1***Other 0.6 1.5***
Degree Received ( %)
High School Diplom_ 48.0 37.9***GED 7.8 8.0None 44.1 54.14,c*
Average Highest Grade Completed 10.9 10.1***
Marital Status (%)Never Married 30.1 11.0***Married, Living with Spouse 5.9 84.8***Married, Not Living witn Spouse 27.6 2.5***Widowed or Divorced 36.5 1.7***
Any Children (%)b
Less Than 6 Years 10.0 72.3***Between 6 and 18 Years 90.4 5/.5***
Mandatory AFDC With Child LessThan 6 (%)c 5.3 0.9***
(continued)
-32-
75"
TABLE 2.3 (continued)
Characteristic AFOC AFDC -U
Monolingual in a Language Otherthan English (%)SpanishOther
Undocumented Worker (%)
Activities Within 12 Months Priorto Initiol Registration (%)
8.4
0.5
0.8
15.6***1.2**
5.8***
Job Search Workshop 16.6 15.0EWEP 9.9 7.9**Education or Training 22.2 15.0***No Prior Activities 63.0 71.2***
Current Activities (%)Employed 20 Hours or Less
Per Week 7.0 6.3Employed 21-30 Hours Per Week 5.5 3.1***
Education or Training 14.6 9.6***
Prior AFDC Dependency (%)Never on AFDC 11.4 34.1***1-11 Months 6.9 15.4***12-23 Months 6.8 10.6***24-35 Months 8.1 10.7***36-47 Months 8.3 7.9
48-59 Months 6.9 6.860 Months or More 51.4 14.5***
Average Number of Months Everon AFDC 69 5 24.4***
Average Number of Months on AFDCDuring 24 Months Prior to initiG1Registration 15.5 9.5***
Ever Included on Someone Else'sAFDC Case (%) 16.7 33.4***
Length of Time Employed During24 Months Prior to InitialRegistration (%)Not Employed 49.9 27.7***1 Week to 6 Months 18.0 19.1
7-12 Months 12.8 18.0***13-18 Months 7.7 13.0***19-24 Months 11.5 22.2***
(continued)
-33-
TABLE 2.3 (continued)
Cha.acteristic AFDC AFDCU
Held a Sob at Any Time DuringQuarter Prior to InitialRegistration (%)e 26.8 38.1***
Held a Job at Any Time DuringFour Quarters Prior to InitialRegistration (x) 39.4 56.5***
Held a Job at Any Time DuringTen Quarters Prior to InitialRegistration (%) 51.7 69.1***
Estimated Earnings During 24Months Prior to InitialRegistration (%)
$0 49.9 27.5***$1$1000 14.0 11.6**
$1001$5000 17.7 21.2***$5001$10,000 10.2 18.2***Over $10,000 8.2 21.6***
Average Earnings During Quarter($)ePrior to Initial Registration ($) 421.85 870.71***
Average Earnings During FourQuarters Prior to InitialRegistration ($)e 1668.60 3507.07***
Average Earnings During TenQuarters Prior to InitialRegistration (S)e 4035.16 8055.14***
Received Unemployment CompensationDuring Three Months Prior toInitial Registration (%)e 4.2 9.2***
Received Unemployment CompensationDuring 12 Months Prior toInitial Registration (%)e 7.5 17.6***
Average Amount of UnemploymentCompensation During Three MonthsPrior to Initial Registration ($)e 32.05 68.79***
Avertige Amount of Unemployment
Compensation During 12 MonthsPrior to Initial Registration Me 126.62 299.75***
Sample Sizef
3227 1387
(continued)
34-
TABLE 2.3 'continued)
SOURCE: MDRC ClieO Information Sheets and the State of CaliforniaUnemployment Insurance earnings and benefits records.
NOTES: The mple far this table includes individuals whoregistered between July 1985 and June 1 °86.
Distributians may not add to 100.0 percent due to rounding.
A chi-square test or t-test was applied to differencesbetween assistance categories. Statistical significance levels areIndicated as: = 10 percent; = 5 percent; = 1 percent.
aAFDC-U cases can be redetermined as WIN-mandatory when an
AFDC case becomes an AFDC-U case or when a Previously exempt AFDC-U case(e.g., medically exempt) loses its exemption status.
bDistributions may not add to 100.0 percent because sample
members can have children in more than one category. In addition, someindividuals, who are not part of their parents' case, may not hake anychildren.
cA few AFDC-B's may be included in the 'Mandatory AFDC With
Child Less Than 6' category due to data entry errors or misinterpretationof the question.
dDistributions odd to more than 100.0 percent because sample
members can be included in more than one activity.
eThese data are calculated from the State of California
Unemployment Insurance earnings records and include .ero values for samplemembers not employed and for those not receiving Unemployment Compensati3n.
f
For selected characteristics, sample sizes may vary up to 5
sample points due to missing data. 62 of these registronts were excludedfrom the impact analysis because they did not nave social security numbers.
. -35 -
reported not working in the two years prior to registration, compared with
50 percent of the AFDC's. The average earnings for the AFDC-U's during the
year prior to random assignment was $3,507, compared with $1,669 for
AFDC's. (Note that these averages include zero values for sample members
not employed.)
Fewer AFDC-U's reported recently engaging in activities aimed at
improving their employability than did AFDC's. As of registration, 15
percent of the AFDC's reported current participation in an education or
skills training activity. During the year prior tL registration, 17 per-
cent of the AFDC's had participated in job zearch workshops; 10 percent had
been active in EWEP; and 22 percent had participated in an education or
skills training program. Among the AFDC-U's, 10 percent reported current
participation in an education or skills training activity. During the year
prior to registration, 15 percent had participated in job search workshops;
8 percent had been active in EREP; and 15 percent had participated in an
education or skills training program. Because of the many differences
between the two assistance groups, the majority of this report will analyze
AFDC-U's and AFDC's separately.
A. Characteristics of Controls and Experimentals
Random assignment to the impact sample proceeded smoothly, resulting
in experimental and control groups with similar demographic character-
istics. Appendix Table B.1 presents the demographic characteristics of the
experimentals and controls in the impact sample. There were only a few
statistically significant demographic differences between the groups at
random assignment: Slightly more AFDC controls than experimentals were
redetermined recipients. More AFDC controls than experimentals were Asians
-36-
and Pacific Islanders.
There were even fewer significant control-experimental differences
among the AFDC-U registrants. (See Appendix Table B.1). A slightly lower
proportion of controls than experimentals reported being employed 21 to 30
hours a week as of registration. Average earnings one year prior to
registration were also lower for controls.
B. Subgroup Characteristics
In addition to estimating overall impacts, the research addr .s the
important issue of whether certain subgroups of individuals are likely to
benefit more fran the SWIM model than ether subgroups. The impact and the
process analyses thus focus on several important subgroups. The primary
division is between the AFDC's and the AFDC-U's.
these two groups have already been described.
A second division is between earlier and later registrants. Given
t at tne members of the impact sample entered the zample corer a one-year
period, it is important to determine if sample members who were randomly
assigned in the later part of the registration period differed syste-
matically fran those who were randomly assigned earls 'Jr. Appendix Table
B.2 shows characteristics of the subgroups registering with SWIM between
July 1985 and December 1985 and those registering between January 1986 and
June 1986.
For the AFDC registrants, the earlier sample is approximately 55
percent of all registrants in the impact sample and the later sample is 45
percent. There are some statistically significant differences between
these two groups. AFDC registrants in the earlier group are more likely to
be applicants and less likely to be renewed recipients than the later
The characteristics of
80-37-
group. This, in turn, itlans that the earlier sample is less disadvantaged
with respect to prior well' :e dependency than the later one, although these
differences are not very large.
These sample differences may be due to the fact that delays in
notifying recipients of their renewal interviews slowed the rate at which
this group was phased into the program. This, in turn, affects the
composition of the sample with respect to the applicant/recipient
distinction. They may also be due, to a lesser extent, to a d'cline in the
unemployment rate in San Diego county over the intake period for the impact
sample; the unemployment rate decreased from 5.3 percent in the last 6
months of 1985 to 4.9 in the first 6 months of 1986. As the unemployment
rate declines, the more advantaged are likely to find jobs and therefore
are loss likely to apply for welfare.
For the AFDC-U registrants, the earlier group is approximately 54
percent of all AFDC-U registrants and the later group is 46 percent.
Comparisons of demographic characteristics reveal few differences between
the groups. The percentage of females was greater in the earlier AFDC-U
sample than the later one; and the employment as well as the average
earnings of the earlier sample were lower than those of the later one. The
AFDC-U samples did not differ, however, along the measures of prior welfare
dependency. It is notewort.iy that, unlike the AFDC samples, the two AFDC-U
groups did not differ in the percentages of akpltcants versus recipients.
It is unclear why notification delays and the declining unemployment rate
did not affect AFDC-U's in the way they seem to have affected AFDC's.
A third important division is between the applicant and recipient
samples. (See Table 2.4). Demographically, these two groups were very
TABLE 2.4
SW!M
SELECTED CHAR...CTERIST1CS OF REGISTRANTS AT THE TIME OF INITIAL REGISTRATION,BY ASSISTANCE CATEGORY AND WELFARE STATUS
Characteristic
AFDC AFDC -U
Applicant Recipient Applicant Recipient
Office (%)
Service CenterSan Diego West
AFDC Status (%)ApplicantRenewed RecipientRedetermined Recipient
Average Age (Years)
Sex (%)
44.9
55.1
100.0
0.0
0.0
33.9
53.1***46.9***
0.0***54.1***
45.9***
34.3
50.1
49.9
100.0
0.0
0.0
31.0
50.7
49.3
0.0***57.5***
42.5***
35.6***
Male 12.9 ).1**4, 92.4 89.6*Female 87.1 93.9*** 7.6 10.4*
Ethnicity (%)White, Non-Hispanic 31.3 24.6*** 29.6 17.6***Black, Non-Hispanic 41.2 42.8 22.3 16.8**Hispanic 21.8 28..'/*** 40.3 44.8American Indian /Alaskan Native 1.0 0.3** 0.4 0.5
Asian and Pacific Islander 3.9 3.6 6.3 18.3***Other 0.7 0.5 1.2 2.0
Degree Received (%)High School Diploma 52.6 45.1*** 42.9 30.5***GED 8.4 7.5 9.8 5.4***None 39.1 47.4*** 47.3 64.2***
Average HighAst Grade Completed 11.2 10.7*** 10.6 9.3,0,0,0
Marital Status (%)Never Married 25.6 33.0*** 12.0 8.3**Married, Living with Spouse 8.3 4.3*** 82.3 88.5***Married, Not Living with Spouse 31.2 25.3*** 2.8 2.2Widowed or Divorced 35.0 37.4 2.2 1.1
Any Children (%)b
Less Than 6 Years 7.0 12.0*** 75.9 67.0***Between 6 and 18 Years 91.0 89.9 49.7 69.0***
Mandatory AFDC With ChildLess Thar: 6 (%)c 2.8 6.9*** 1.0 0.9
-39-82
(continued)
TABLE 2.4 (continued)
Cherocteristic
AFDCr
AFDC-U
Applicont Recipient Applicant Recipient
Monolingual in o LonguogeOther Thon English (t)Spanish 7.0 9.2** 11.7 21.3***Other 0.9 0.3** 0.4 2.5***
Undocumented worker (%) 0.9 0.8 6.0 5.6
Activities Within 12 Months Priorto Initidl Registration (%)Job Search Workshop 10.7 20.5*** 9.3 23.5***EWEP 5.2 13.0*** 3.5 14.3***Education or Troining 16.0 26.1*** 10.! 22.2***No Prior Activities 73.7 56.0*** 81.2 56.3***
Current Activities (X)Employed 20 Hours or Less 6.5 7.3 3.6 10.4***
Per Week
Employed 21-30 Hours Per Week 4.3 6.2** 2.8 3.6Education or Troining 9.7 17.7*** 5.9 15.1***
Prior AFDC Dependency (%)Never on AFDC
e22.3 4.3*** 52.0 7.5***
1-11 Months 12.8 3.2*** 20.0 8.6m12-23 Months 9.1 5.4*** 8.7 13.4***24-35 Months 9.1 7.4 6.6 16.8***36-47 Months 7.6 8.8 3.5 14.3***48-59 Months 6.3 7.3 3.1 12.2***60 Months or More 32.8 63.5*** 6.0 27.1***
Averoge Number of Months 47.6 83.6*** 12.0 42.8***Ever on AFDC
Average Number of Months on AFDCDuring 24 Months Prior toInitial -Registration 8.0 20.4*** 3.7 18.1***
Ever included on Someone Else'sAFDC Case (%) 15.9 17.2 32.6 34.6
Length of Time Employed During24 Months Prior to initiolRegistration (S)Not Employed 33.9 ,(0.3*** 9.9 54.1***1 Week to 6 Months 18.3 17.9 18,7 19.77-12 Months 15.3 11.2*** 21.1 13.3***13-18 Months 12.9 4.2*** 18.2 5.4***19-24 Months U.S 6.3*** 32.1 7 .5 .s
-40- 83
(continued)
TABLE 2.4 (continued)
Characteristic
AFDC
Applicant Recipient
Held o Job of Any Time DuringQuarter Prior to Initial
Registration (%)
Held o Job of Any Time During FourQuarters Prior t? InitiolRegistration (%)
38.6
50.5
19.1 ***
32.3***
Hold o Job of Any Time During TenQuarters Prior t? Initial
RegistratiAn (s) E9 .6 46.6***
Estimated Earnings During 24 MonthsPrior to !nitiol Registration (%)
$0 33.9 60.2***$1 - $1,000 12.9 14.8
$1,001 - $5,000 19.4 16.5**$5,001 - $10,000 16.2 6.3***Over $10,000 17.6 2.1***
A/eroge Earnings During Quor;erPrior to Initial Registration (5) 764.67 201.03***
Average Earnings During FourQuarters Prior t? InitialRegistration (5) 2993.98 B16.91***
Average Earnings During TenQuarters Prior ? Initial
Registration ($) 6924.97 2178.16***
Received Unemployment Compensation-During Three -Mon/hs Prior to Initial
Registration (%) 7.8 1.8t**
Received Unemployment CompensotionDuring 12 Months Prior t?
Registration (%) 12.5 4.3***
Average Amount of UnemploymentCompensation During Three MonthsPrior to initial Registration ($)t 63.39 11.86***
9.9
10.0
22.9
23.8
33.4
1286.52
5182.35
11,271.59
13.2
23.8
100.40
53.6***14.n**18.6*
3.9,0,0*
260.39***
1057.39***
3349.14***
3.3101,N
8.5***
22.33***
(continued)
-41-
TABLE 2.4 (continued)
Characteristic
AFDC AFDC-U
Applicant Recipient Applicant Recipient
Average Amount of UnemploymentCompensation During 12 MonthsPrior to Initial Registration ($) f
237.54 55.17*** 403.83 146.78***
Sumpie Size 91267 1960 829 558
SOURCE: See Table 2.3.
PJTES: The sample for this table includes Individuals who registered betweenJuly 1985 and June 1986.
Distributions may not add to 100.0 percent due to rounding.
A chi-square test or t-test was applied to differences between welfarestatuses within assistance categories. Statistical significance levels are indicatedas: * = 10 percent; ** = 5 percent, *** = 1 percent.
aAFDC-U cases can be redetermined as WIN-mancictory when an AFDC case
becomes on AFDC-U case or when o previously exempt AFDC-U case (e.g., medicallyexempt) loses Its exemption status.
bDistributions may not add to 100.0 percent because sample embers con
hove children in more than one category. !n addition, some individuals, who are notpart of their parents' case, may not have any children.
cA few AFDC-U's may be included in the 'Mandatory AFDC With Child Less
Than 68 category due to data entry errors or misinterpretation of the question.
dDistribution' add to more than 100.0 percent because sample members
can be included in mare than one (ctivity.
eA few recipients may be included in the 'Never on AFDC' category due
to data entry errors or misinterpretation of the question.
(These data are calculated from the State of California UnemploymentInsurance earnings records and include zero values 7or sample members not employedand for those not receiving Unemployment Compensation.
9For salected characteristics, sample sizes may vary up to 5 sample
points due to missing data. 62 of these registrants were excluded from the impactanalysis because they did not have social security numbers.
4286
different, as would be expected since sane individuals who have never
received welfare or whose appl cations will be denied are included among
the applicant group. Applicants had more recent work experience.
Fifty-one percent of the AFDC applicants had held a job during the year
prior to random assignment compared with 32 percent of the AFDC recipients.
Sixty-eight percent of the AFDC- -U applicants had a job during the year
prior to random assignment, compared with 39 percent of the AFDC-U
recipients. Applicants also tended to be more educated. Fifty-three
percent of AFDC and 43 percent of AFDC -U applicants had a high school
diploma as compared with 45 percent of the AFDC and 31 percent of the
AFDC-U recipients. Applicants, as expected, had less histo:y of welfare
dependency than recipients. However, it is Important to note that, even
among applicants, only 22 percent of the AFDC's and 52 percent of the
AFDC-U's had no welfare history at all.
A fourth division is between registrants at the two welfare offices
invoived in the demonstration. Demographically, these two groups were
generally similar. However, a higher proporticon of registrants at the San
Diego West office were applicants, as compared to registrants at the
Service Center office. As a result, registrants at the San Diego West
office were better educated, had less history of welfare dependency, and
more recent ties to the labor force.
IV. Sample Used to Measure Monthly Participation
As mentioned, the current WIN mandatory caseload was phased into the
SWIM program during its first year of operation. The one-third who were
randomly assigned to control group status, of course, were not eligible to
-43-
receive SWIM services. During the second year of SWIM all new regt_crants
-- primarily new applicants and recipients recently determined to be
WIN-mandatory -- were eligible to receive servtees. Thus, no registrants
were placed in a control group. To be able to study the extent to which
saturation was reached over an extende., period of time, data were collected
for a representative sample of approxit,ately 33 percent of the individuals
registering with SWIM during the year after the End of random assignment,
i.e., between the period of July 1, 1986, through the end of June 1987.
Data on this sample of registrants were combined with data on the
SWIMHeligibles in the impact sample to analyze case management and parti-
cipation patterns for the entire caseload, including both AFDC's and
AFDC-U's. To approximate the size of the entire caseload, several weight-
ing factors were used. The combined AFDC and AFDC-9 sample was weighted to
adjust for the higher proportion of AFDC's, relative to the proportion of
the AFDC-U's, for whom data were collected. The sample was also weighted
to adjust for the fact that data were collected for only a portion of the
group.5 Weighting in this way enables the monthly partici-
pation analysis to draw conclusions for the estimated size of the actual
caseload with which the staff wa working rather than the smaller samples
used in other parts of the research.
Those who registered with the program during its first year of
operation (i.e., members of the impact sample) had demographic character-
istics quite different from those who registered wfth the program after
June 1586. (See Appendix Table B.3.) This is to be expected, since all
the current WIN-mandatory registrants were phased into the SWIM program
durthg the first year. These individuals were already on welfare and many
had a long-term history of welfare dependence. After the ft:1-st year of
SWIM operation all the current WIN-mandatory registrants had already
registered for SWIM. Those who registered for SWIM in the second yeal
tended to be applicants or redetermined recipients. 6 Only 40 percent of
AFDC's who registered before June 1986 were applicants, compared to 58
percent of the post-June 1986 AFDC registrants. Both the AFDC's and
AFDC-U's who registered with the program after June 1986 had less welfare
dependency than their countc,parts who registered before June' 1986.
V. Other Research Samples
In addition to the main research questions discussed in Chapter 1, the
research addressed several secondary questions that required looking at
specific aspects of SKIN in greater detail. To examine these secondary
research questions, four "Hall random subsamples were selected from amon
the SWIM-eligibles.
A. Childcare SamplJ
To analyze the use of childcarf:. in SWIM the case file records of 121
registrants were reviewed. This sample of 121 registrants was randomly
selected from AFDC SWIM - eligibles registered and participating in SWIM
components during July or November of 1986.
B. Nonparticipation Sample
One of the primary research questions in SWIM, as noted, is whether 75
percent of the registrants in any given month will participate in at least
one compoi,lnt. A special sample of 99 registrants was randomly selected
from those registrants not participating during July or November of 1986 to
discover reasons for nonparticipation through case file reviews.
-45- 88
C. Noncompliance Sample
To describe the extent to which individuals were nmt in compliance
with program requirements, case file records of a sample of program
registrants, both participants and nonparticipants, were reviewed. The
sample consisted of 144 AFDC and 98 AFDC-U registrants, randomly selected
from individuals who were randomly assigned to the SWIM-eligible research
group between January 1 and March 31, 1986. These registrants' program
activities were tracked for 15 to 18 months after random assignment.
D. Worksite Sample
Worksite interviews were conducted by MDRC with a subsample of 30 work
experience supervisors, primarily to obtain estimates of the value of work
done by EWEP participants. Thirty registrants were randomly selected, from
the May 1937 EWEP assignment logs, from among the 85 registrants who were
assigned to participate in May 1987. The supervisors of these 30
registrants were interviewed for about 20 minutes over the telephone.
VI. Data Sources
Ti-Il report uses a number of different data sources to analyze the
flow of 2r Widuals through the prograr', to describe program operations and
implementation, and to measure employment and wUfare outccines. As
indicated in Table 2.5, these sources provide varying lengths of follow-up,
depending on the sample member's initial registration date. The sources
are:
Client Information Sheet (CIS) is a one-page interviewdocument designed by MDRC to provide data on registrants'demographic characteristics such as age, ethnicity, familycomposition and education and training history, as well asinformation on their welfare and employment histories. This
-46-
TABLE 2.5
SWIM
LENGTH OF AVAILABLE FOLL 11-UP BY DATA SOURCE AND PERIOD OF INITIAL REGISTRATION
Data Data Source
Point
at Which
Data
Collection
Begins
Lost Date Data
Are Available
Length of Follow -Up By Period of initial Registration
July-
September
1985
October-
uocember
!985
January-
March
1986
April-
June
1986a
Process Data 9NIA Automated Tracking Date of June 1987 Twenty -One Eighteen Fifteen TwelveSystem and EINEP Attendance initial Months Monthsb Months
bMonths
Lags° Registration
Quarterly Employment State of California 10 Quarters First Quarter Six Quartersd
Five Four Threeand Earnings Data Unemployment Insurance Prior to 1987 Quarters
dQuarters
dQuarters
System' Initial
Registration
Monthly AFDC County of Son Diego 20 Months August 1951 Twenty-Four Twenty-One Eighteen FifteenGrant Payments AFDC Paymer4s System Prior to
initial
Monthse Monthse Monthse
Monthse
Registration
NOTES:aTracking data was not collected for members of the control group.
bThe first month of fallow -up for tracking data does not include the month in which an individual Initially registered.
cUnemployr9nt Insurance earnings records report earnings on a calendar quarter basis.
dThe calendar quarter of initial registration is not considered to be a follow-up quarter for employment and earnings for the
981M evaluation.
coa,
The first month of folio/I-up for AFDC grant payments includes the month in which on individual initially registered.
form was completed by SWIM program staff for registrants atthe time of SWIM registration. These data were then mergedwith :Iformation on welfare 2ceipt, employment and programparticipation in the final analysis file. Data quality wasgenerally good.?
California State Unemployment Insurance (UI) Earnings andBenefits Records provide measures of earnings reported bycalendar quarter: i.e., January through March; April throughJune. Unemployment benefits data are reported by calendarmonths to coincide with the payment schedule of thesebenefits.
Several limitations of these data should be noted. First,because of the reporting lags typical of the UI wage reportingsystem, data were only available for three quarters afterrandom assignment for the entire impact sample.' Second, theuse of quarterly data meant that there were varying lengths offollow-up, depending on whether an individual registered forSWIM during the first, second or third mont of thecalendar quarter. Third, even for existing data, there couldbe some underreporting -- for example, because of employersfailing to report earnings or people moving out of state.Also, not all employers are required to report. Thus, UI datado not necessaLily cover all employment of the researchsample. Since all these factors should have affected experi-mental and control group members equally, there is no reasonto believe they affected employment and earnings outcomesdifferently fol: experiraentals relative to those of controls.9
AFDC Records supply information on monthly AFDC (i.e.,welfare) grants. These data were obtained directly from theCounty of San Diego Department of Social Services andcollected through August 1987 for the analyses is th.s report.This provided 2.5 months of post-random assignment follow-upfor the entire impact sample. When AFDC data are matched toCIS and UI data, sane inaccuracies, due either to incompletedata entry or inability to match records, can be expected.Since this source of error should not differ across researchgroups, it should not be a sourle of bias for the impactestimates. 10 In order to be c; latible with the earningsdata, welfare payments were aggre, .ed into calendar quarterperiods.
The SWIM Automated Tracking System was used as a case manage-ment system by SWIM program staff as well as to provide datafor the research. This system was used to provide informationon program registration and deregistration, as well as startand end dates for program-related activities such as jol.search workshops, job clubs, and program-arranged educationand training activities; and for registrant-initiated
-48-
92
activities such as employment, or education or trainingprograms in which registrants enroll on their own.11 Thesystem maintained data on all SWIM - eligibles. Data wereprocessed through June 1987, providing 12 months of follow-upfor SWIM - eligibles i, the impact sample. MDRC conducted acomprehensive quality check of the SWIM tracking system. Theresults indicated that, for the most part, the automatedtracking system provided adequate data for 'salysis. (SeeAppendix A for more detail.)
EWEP Lan maintained by the San Diego Workfare Unit within theDSS Employment Services Bureau were used to provide inform-ation on worksite attendance. The logs werc4 completed by theEWEP staff at each of the local welfare offices andperiodically sent to MDRC. These data were collected throughJune 1987 for all SWIM registrants, providing 12 months offollow-up for the entire impact sample.
Interviews with program staff and education and trainingproviders were used in addition to direct observation ofprogram activities and review of local office case files tostudy program activities for the process analysis.
CHAPTER 3
ADMINISTRATION AND STAFFING
To provide a context for understandin the participation and impact
findings presented later in the report, this chapter describes the
administration and staffing of the SWIM program, including coordination
between staffing units, case management procedures and staff attitudes
regarding the program. Chapter 4 then describes the nature of program
services.
I. Administrative Structure
The administrative structure of SWIM was complicated, as indicated in
Figure 3.1. Building on the EPP/EWEP program model, the program was
administered by the County and State Departments of Social Services (DSS)
and the State Employment Development Departmen.: :YDD) and its district
offices. The county's lead agency for the demonstn .n was the Employment
Services Bureau within DSS, which coordinated work San Diego's EDD.
Within the Employment Services Bureau, the Employment *x=tion Division
provided overall direction for pro4ram registration, entation, job
search workshops, job clubs and the SkUls Techniques Achievement Reviews
(STAR); the Workfare Division directed the EWEP program; the SWIM Division
directed assessments and education and training activities; and the
Employment Training Division directed Individualized Supervised Employment
Search Activity (ISESA). In addition, education and training services were
provided by community organizations. Thus, actual services were provided
State Employment
Development Dopartment
(EDD)
County of San Diego
.) District VII
EDD District Office
employment
laists
Job Search
Workshops
and Job Clubs
95
FIGURE 3.1
ORGANIZATIONS, STAFF AND SERVICES IN THE SWIM DELIVERY SYSTEM
EEmployment
Services Bureau
. _
State Department
of Social Services(DSS)
County of San Diego
Department of
Social Services.,------..----
Empioyment
operation Di . $
Social Workers
Registration,
Orientation
and Appraisal
Workfare Division
I
Job Devek,.per j Job Developer j Job DeveloperCounselors I Counselors (CRU / Counselors
1
SWIM Divisionll
inployment TrDivision
[Job DevelopmenlUnit
EWEPAssessment,
EducationEducation andTraining Referrals
ISESA
Income Maintenance
Bureau
ome Maintenanc
District Offices
Technicians
Determination
Of WelfareBenefits
Adult School,
Community Cc",9eand JTPA Programs
raEducation
and Training
Services
9 6
by several different agency or unit staffs.
II. Staffing Levels and Staff Responsibilities
Managing the flaw of registrants through the program as well as
providing direct program services required a large staff. This was true in
spite of the fact that the county welfare department referred program
registrants to cammunity resources for education and training and did not
operate or fund these activities itself. Substantial staff resources were
required to monitor attendance, deal with noncompliance with program
requirements, arrange support services and track registrants' activities.
In part, this was due to the fact that the SWIM autanated tracking system
was not exploited to the fullest extent possible to aid in case management
tasks. Program resources were not available to develop computer routines
that would all the tracking system to do same of the more clerical work
^r_rrormed by the staff who were responsible for keeping detailed records of
registrants' program activities.
Building on procedures used in the EPP/EWEP program, different sets of
staff were responsible for working with and monitoring registrants as they
progressed through different stages of the program model. This section
uses the results of an MDRC time study to indicate the program functions
fulfilled by each set of line staff involved in SWIM.
The discussion begins by summarizing the overall findings of the time
study for professional staff involved in SWIM. It then provides more
detail on the responsibilities of both professional and clerical staff.
-52-
97
A. Overview of Activities of Professional Staff in SWIM
The results of MDRC's time study, which was conducted in June 1986,
are shown in Table 3.1. Note that in addition to the staff indicated in
Table 3.1 and described below, supervisory staff (located at each SWIM
office) and administrative staff (loca,ed at a central office) were part of
the program's staffing configuration.
Regardless of program component, two-thirds of professional staff time
was spent monitoring attendance, dealing with noncompliance, arranging
support services and tracking registrants' activities. In particular,
arranging and authorizing support service payments was very time-consuming;
over one-quarter of all staff time was spent on this activity.
Broken down by component, twelve percent of SWIM professional staff
hours were spent on activities that occurreC during the morning sessions
when individuals registered with the program. These activities included
random assignment, completion of a one-page research document which
recorded registrants' background characteristics, registration, program
orientation and post-orientation appraisals.
Over 45 percent of professional staff hours were occupied by tasks
associated with providing job search activities. This included arranging
for_support services for participants, leading the job search workshops or
job clubs, and dealing with noncompliance associated with these activities.
Seventeen percent of professional staff hours were spe& in connection
with EWEP. EWEP activities included orienting registrants to EWEP, placing
registrants in EWEP and monitoring their participation, arranging support
services for participants, dealing with noncompliance in EWEP and
miscellaneous EWEP program reporting.
-53-9u
93-
TABLE 3.1
91
PERCENTAGE OISTRIBUTION OF ALL NON-SUPERVISORY ANO NON-ADMINISTRATIVE STAFF TIME SPENT ON
SWIM TASKS, BY TASK, AGENCY, ANO TYPE OF STAFF
SWIM Task
EPP E00 EWEP SW IAA TOTAL
SocialWorkers CI erks
Employ-ment
Special-ists CI erks
JobCleve' c-
perCounse-I ors CI erks
JobOevel o-per
Counse-I ors Clerks
Pro-tesslon-
alStaf f CI erks
Random Assignment and CompletingCl lent Information Sheet 0.0 0.0 0.0 0.0 0.0 25.4 9.5 23.1 2.1 7.4
Registration and Orientation forSel IM-EI I gi bl es and Control s 6.8 83.9 3.4 13.9 0.0 0.0 0.0 4.5 3.5 44.6
Appraisal s for SHIM-El igi bi es;Exit Conferences for Contra( s 15.2 13.0 1.6 0.0 0.0 0.0 0.0 0.0 5.9 6.4
Support Service Arrangements for SHIM -Eligibles in Job Search Workshopor Jot Club 19.0 0.0 27.4 66.9 0.0 0.0 0.0 0.0 15.6 13.9
Operating Job Search Workshop 24.6 0.0 20.3 0.0 0.0 0.0 0.0 0.0 15.2 0.0Operating Job Club 0.0 0.0 20.2 7.2 0.0 0.0 0.0 0.0 6.6 1.5Non-Ccapi lance Activities Associated with
Registration, Job Search Workshop orJob Club 2.3 0.0 21.6 0.0 0.0 0.0 0.0 0.0 7.9 0.0
Orientation to ENEP, Piecing SWIM - Eligiblesin ENEP, Monitoring Works( tes andBVEP Participants 0.0 0.0 0.0 0.0 44.1 0.0 0.0 0.0 4.7 0.0
Support Service Arrangements forParticipants in ENEP 18.4 0.0 0.0 0.0 17.0 0.0 0.0 0,0 8.3 0.0
Program Reporting for BIEP . 0.0 1.9 0.0 0.0 27.1 74.6 0.0 0.0 2.9 14.3Non-Compi lance Follow -Up for ENEP 0.0 0.0 0.0 0.0 11.8 0.0 0.0 0.0 1.3 0.0 j
( con ti nued)
TABLE 3.1 (continued)
SWIM Task
EPP E00 MEP SWIM TOTAL
Social
Workers Clerks
Employ-
ment
Special-
ists Clerks
Job
Oevelo-
per
Counse-
lors Clerks
Job
Develo-
per
Counse-
lors Clerks
Pro-
fession-
of
Staff Clerks
Assessment for Referrals to Education or
Training, and Monitoring of Participants 0.0 0.0 0.0 0.0 0.0 0.0 24.6 6.6 5.3 0.8Support Service Arrangements for
SWIM-Eligibles in Program-Arranged
Education or Training 4.4 0.0 0.0 0.0 0.0 0.0 0.0 0.0 1.5 0.0Miscellaneous Paperwork for Education
or Training 0.0 0.0 0.0 0.0 0.0 0.0 3.4 2.8 0.7 0.3Non-Compliance Activities Associated
with Educatio: it Training 1.4 0.3 0.0 0.0 0.0 0.0 1.2 0.0 0.7 0.1
Verification of Attendance in Self-lnitioted
Activities and Employment; Tracking
Other Program Statuses° 0.0 0.0 0.0 0.0 0.0 0.0 37.1 63.0 8.0 7.7General Paperwork°2.6 0.0 5.3 6.1 0.0 0.0 24.2 0.0 7.9 1.3
Support Service Arrangements for Controls 4.7 0.0 0.2 0.3 0.0 0.0 0.0 0.0 1.7 0.1Services to Volunteers or Non-Federols 0.5 0.9 0.0 5.6 0.0 0.0 0.0 0.0 0.2 1.6
Total100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0
Number of Staff Surveyedc19 9 15 3 12 6 10 4 56 22
SOURCE: Calculations from MDRC Time Study of SWIM staff, conducted from June 16 to June k7, 1986.
NOTES: Distributions may not odd to 100.0 percent due to rounding.
All percentages are coWloted as a proportion of the total number of staff hours.
Tots of statistical significance were not examined.
101
(continued)
1 0 2
TABLE 3.1 (continued)
°Trackingother program statuses includes investigating SWIM -El 191 bl es' job seorch status, ENEP status, or AFDC/WIN status.
bGeneral poperwork includes completing status change forms, deregistrotion forms, tracking system forms, state reporting
forms, error corrections and miscellaneous poperwork.
cTwo of the EDD Employment Specialists were port-time. In addition, BVEP staff worked with registrants from non-SWIM
offices as well as 9/11M-E1 igibl e registrants.
l
104
Eight percent of professional staff hours were spent in connection
with referring registrants to community education or training programs.
This included assessing registrants for appropriate 'next steps* in the
program, referring individuals to education and training programs and
monitoring their participation, arranginfl support services for parti-
cipants, dealing with noncompliance in these activities and miscellaneous
paperwork associated with program - arranged education and training.
Another 8 percent of professional staff hours were spent verifying the
attendance of registrants in self-initiated education, training or employ-
ment. An additional eight percent of staff time was spent doing general
paperwork, that is, completing status change forms, deregistration forms,
automated tracking system forms, state reporting forms, and doing error
correction and miscellaneous paperwork.
Finally, 2 percent of professional staff time was occupied by
arranging support services for members of the control group.
B. Employment Preparation Division Staff
As of June 1986, 18 Employment Preparation Division social workers and
9 clerks staffed the SWIM program, a staffing level that remained the same
throughout the demonstration. 1Most of these social workers were not new
to their positions and their job duties generally did not change when the
EPP/EWEP program evolved into SWIM.
Employment Preparation Division staff were involved in a variety of
activities. The largest share of the social workers' time -- 42 percent --
was spent arranging, authorizing and supervising support services for
SWIM eligible registrants throughout their program tenure. These support
services included childcare and transportation for those in SWIM activi-
-57- I
ties and work expense advances ("entered employment' stipends) for those
who found jobs. This advance money (usually $50) was intended to defray
work expense costs until the registrant received his/her first paycheck.
Another primary job duty, which occupied 25 percent of the social
workers' time, was assisting with job search workshops. This component was
jointly staffed and supervised by Employment Preparation Division staff and
EDD staff, with EDD taking the lead.
Social workers were also responsible for registration, program orienta-
tion and post-orientation appraisals for both SKIM-eligible and control
group registrants. Registration and orientation occupied 7 percent of the
social workers' time; 15 percent of their time was spent on the appraisal
process.
A small amount of the social workers' time -- 5 percent -- was spent
providing support services to registrants in the control group. Although
controls were not eligible to participate in SKIM components, both
SWAM- eligibles and controls could qualify for the entered employment
stipends.
Finally, 4 percent of social workers' time was occupied by noncompli-
'ance follow-up, conciliation and formal adjudication proceedings for
individuals who did not cooperate with the registration, job search work-
shop, job club or education or training requirements.
Employment Preparation Division clerks assisted primarily with the
paperwork associated with registration, orientation and appraisals.
C. EDD Staff
As of June 1986, 15 EDD employment specialists and 3 EDD clerks
staffed the SWIM program. EDD staffing levels also generally remained the
same throughout the demonstration.
The primary job function of the EDD employment specialists consisted
of leading job search workshops, job clubs and, later, STAR. Approximately
41 percent of these individuals' time -- split evenly between workshops and
job clubs as of June 1986 -- was spent on this type of work. As was the
case with the social work staff, many of the EDD staff had extensive
experience leading job search workshops prior to SWIM, that is, in the
EPP/EWEP program. Operating job clubs and STAR, which were not part of the
EPP/EWEP program, were new responsibilities for the EDD staff.
The authorization of support services in connection with job search
workshops and job clubs accounted for over one-quarter of EDD employment
specialists' staff time. Over one-fifth of their time was spent dealing
with noncompliance in these activities.
EDD clerks assisted primarily with the paperwork associated with
assigning individuals to job search activities ('enrolling' individuals in
EDD) and providing support services.
D. EWEP Staff
EgEP staff, who were part of the Workfare Division, were not located
in the office space shared by the other staff units involved in SWIM but
co-located with Income Maintenance (IM) staff. EgEr staff in four IM
offices -- Southeast, South Bay, Northeast and Kearny Mesa -- served SWIM
registrants.
Among these four offices, approximately 12 EWEP job development
counselors and 6 EWEP clerks served SWIM clients. Although the overall
level of staffing remained relatively stable throughout the period in which
SWIM operated, SWIM registrants constituted varying proportions of the
-59- 107
total EWEP caseload in each office, depending on the IM district. (These
offices served registrants from non -SWIM offices as well.) For example,
practically all EWEP referrals to the Southeast IM office were SWIM
registrants; and approximately 90 percent of the SWIM registrants sche:lled
for EWEP orientations were referred to that office. Very few of the
referrals to the Kearney Mesa office were SKIM registrants.
A breakdown of the EWEP staff time accounted for by SWIM registrants
is shown in the Table 3.1. The largest portion of EWEP counselors' time --
44 percent -- was spent conducting EWEP orientations, developing and
monitoring worksites, placing registrants at worksites, monitoring and
counseling EWEP participants, and making initial contact with registrants
who failed to show up at worksites or stopped working.
EWEP program reporting functions accounted for a little more than one-
quarter of the counselors' time. This reporting refers to the preparation
of specific EWEP forms (e.g., the monthly EWEP logs) and miscellaneous
paperwork not associated with EWEP placement, counseling, arrangements for
social services or noncompliance follow -up.
Seventeen percent of staff time was spent on support services arrange-
ments for EWEP participants. This included assessments and counseling
concerning support services and the paperwork associated with authorizing
childcare and transportation payments.
A relatively small proportion of time -- 12 percent -- was spent on
dealing with EWEP noncompliance. This included issuing formal adjudication
forms, conciliation and initiating sanctioning procedures.
The primary function of the ENEP clerks was to assist with all types
of paperwork associated with EWEP. Sane of the central office EWEP clerks
-60-
also assisted with random assignment.
E. SWIM Staff
SWIM Division staff, known as the Coordination and Referral Unit
(CRU), served several functions within the program. These staff members
conducted assessments for registrants who had completed job search and/or
EWEP, made referrals to community education and training providers, and
kept track of the activities of SWIM - eligibles throughout .he program via
the SWIM automated tracking system.
CRU staff consisted of Job Developer Counselors (JDCs) and clerks. In
any given month, ten or 11 JDCs staffed the program. The number of SWIM
clerks staffing the program increased over time, however, from two clerks
in October 1985 to six clerks in August 1986.
Unlike other staff involved in SWIM, CRU staff had little or no
experience with the county's previous work/welfare programs. Many of these
individuals were recruited from outside the county welfare department; and
among those with previous county experience, only a few had been involved
in EPP/EWEP. Many of the CRU staff, however, had worked in same capacity
as job counselors in the San Diego area and had extensive community service
program contacts.
Tracking registrants' activities and monitoring participation were the
predominant job functions of the JDCs. Over one-third of the JDCs' time
was occupied in this manner. These types of activities included time spent
investigating registrants' job search status, EWEP status or welfare/WIN
status, if this information was not clear in the automated tracking system.
Also included was time spent verifying registrants' participation in
activities which had deferred them from regular SWIM activities -- that is
-61- 1 09.
self-initiated education and training -- as well as verifying employment.
About one-quarter of the JDCs' time was occupied by general paperwork
not specifically associated with any of the program components. This
included completion of deregistration forms, SWIM automated tracking system
data entry forms, state reporting forms, tracking system error correction
forms, and miscellaneous paperwork.
Another quarter of the JDCs' time was spent working with registrants
who had reached the education and training portion of the program model.
In this capacity, the JDCs conducted assessment interviews to determine the
suitability of registrants for ISESA or education/training activities,
located various education and training programs, referred registrants to
these programs or to the ISESA component, monitored the participation of
those referred, counseled the participants, and completed all paperwork
associated with these activities.
JDCs also spent a small amount of time interviewing registrants in
order to complete the Client Information Sheet, a research document which
obtained registrant demographic information. This activity took up 10
percent of the JDC's time.
These staff spent minimal time completihg miscellaneous paperwork (3
percent) and dealing with noncompliance in education and training
activities (1 percent).
CRU clerks assisted with the paperwork associated with several types
of activities. As of June 1986, when four CRU clerks staffed the program,
the majority of clerk time was spent assisting with general program tracx-
ing and attendance monitoring duties associated with case management. When
two more clerks were added to the CRU staff, the clerks were able to take
-62-
110
on more of the clerical tasks involved in case management. By the begin-
ning of 1987, CRU clerks had taken over the JDCs' responsibility for
verifying registrants' employment and were providing assistance in verify-
ing the attendance of those in self-initiated education or training.2
During the period of random assignment almost one-quarter of CRU clerk
time was spent randomly assigning registrants to the control or SWIM-
eligible group or checking whether a new registrant had previously been
assigned to one of these groups.3 As of June 1986, a small amount of CRU
clerk time was spent assisting with the monitoring of program-arranged
education and training program attendance (7 percent), assisting with
registration and orientation (5 percent) and miscellaneous paperwork
associated with education and training (3 percent).
F. Employment Training Division Staff
Generally two Job Developer Counselors from the Employment Training
Division led the ISESA workshop component throughout the SWIM demonstra-
tion. Additionally, approximately ten Job Developer Counselors were
involved in developing on-the-job training positions for welfare
recipients. These staff did not work exclusively with SWIM registrants and
were not part of the MDRC time study of staff job functions. However,
interviews with the ISESA workshop leaders indicated that their ISESA job
duties consir,ced of many of the same activities EDD staff performed in
connection with job search workshops: leading the job search sessions,
arranging for support services, and following-up on non-attenders.
III. Coordination Among Staffing Units
When SWIM started, there was sane confusion among staff as to the
-63-.1 1:1_
roles of the various sets of program staff. .°Clder' scaff were initially
unsure of the responsibilities of the CRU staff, who were added to the
staffing configuration which existed in the EPP/EWEP program. Sane of the
social work staff, in particular, viewed the CRU staff as taking over their
role. In time, roles became better defined between the two sets of staff
and coordination improved. It should also be noted that coordination per
se involved little more than making referrals to other staff once a
registrant completed an activity.
IV. Case Management Procedures
Two general models of case management exist, with sane programs using
a hybrid of the two. In same programs, a single staff person has contact
with each registrant to perform assessments and provide support throughout
the registrant's program eligibility. In other programs, registrants are
referred from one set of staff cc, another as they progress through the
program model.
The SWIM case management structure is an example of the second type.
As indicated by the previous discussion, different sets of staff were
responsible for registrants as they progressed through SWIM. Throughout
the program, however, the Employment Preparation Division staff monitored
childcare arrangements, while the CRU staff made sure that all registrant
activity was recorded in the SWIM automated tracking system.
Registrants generally progressed smoothly through the program.
Several factors facilitated this achievement. First, the county had
operated the EPP/EWEP program, using this method of case management, for
three years prior to the start of SWIM. Consequently, many case management
-64112
procedures were firmly established and staff were familiar with them prior
SWIM. Second, the fixed-sequence nature of the pre-assessment portion
of the SWIM program model -- job sear-h followed by EWEP -- made referrals
ern components almost automatic. Until a registrant reached the
assessment stage, the program model afforded staff with little opportunity
for discretionary decisionmaking concerning next steps for the registrant.
As evident from the above discussion, the caseloads of most staff
consisted of all registrants referred to their portion of the program
model. Once registrants completed a specific activity -- for example, job
search workshop -- they would leave the caseload of their workshop leader
and become part of the caseload of an EWEP counselor.
CRU staff caseloads were the exception: Individuals were assigned to
CRU staff caseloads when they registered with the program and remained in
the CRU caseload until they deregistered from the program. For each
individual in their caseload, CRU staff were responsible for ensuring that
the automated program tracking system reflected all activity from
registration through deregistration, regardless of the activities of the
registrants. CRU staff also verified the continuing employment of working
registrants, monitored the attendance of those in self-initiated education
and training, and assessed and referred to education and training programs
any individuals who reached this stage of the model.
Given these responsibilities, CRU caseloads grew over the course of
the demonstration as the WIN-mandatory caseload was progressively phased
into the program. In December 1985, for example, CRU caseloads averaged
approximately 197 registrants.4 Over the following year, CRU caseloads
averaged 214 in February 1986, 250 in April 1986, '494 in June 1986, 331 in
-65- 113
August 1986, and 328 in December 1986. At this point it was clear that CRU
caseloads were too big. Registrants who were working while in the program
were transferred to clerk caseloads, and all individuals in self-initiated
education or training were consolidated in the caseload of a single JDC.
As a result, CRU caseloads dropped to approximately 260 in March 1987 and
246 in June 1987.
For the CRU staff, SWIM case management involved the SWIM automated
tracking system. This system was designed in early 1985 to aid in case
management, as well as to provide data for the research. Data were entered
into the system from individual data input documents completed by the CRU
staff and fran program activity logs (that is, job search workshop, job
club and EWEP attendance logs). The major data items in the system were
activity type, provider codes and dates of activity referrals, starts,
interruptions and completions. Status information -- such as registration
and deregistration dates, and reasons for deregistrations -- were also
recorded in the system. In addition to recording changes in registrants'
program activities or statuses, the system was designed to record
verifications of registrants' continuing activity in order to provide case
management information to program staff.
The reality was different from the design, however. The tracking
system, in fact, functioned primarily as a data depository and not as an
interactive system that could aid in case management. The county did not
have the staff or resources to develop computer routines that would allow
staff to make extensive use of the system for case nagement. For
example, the system did not have a working !tickler! function to provide
staff with lists of registrants in their caseloads who required that sane
-66- 114
action be taken during particular weeks. To identify registrants requiring
follow-up, CRU staff had to scan computer output which listed, in chrono-
logical order, the 20 most recent entries for each registrant in their
caseload. (Examples of entries would be start dates, interrupt dates or
end dates for various activities.) Additionally, the county was unable to
program automated edits -- computer routines which could have identified
missing or incorrect entries to the system. Instead, to determine if all
registrant information was accurately recorded in the system, CRU staff
would periodically review these caseload activity reports and assess
whether the activity entries for each registrant made sense and looked
complete. Staff would also compare information on activity logs to the
information found in the system.
V. Staff Attitudes Toward the Program
Overall, program staff reacted favorably to the SWIM program model.
Among interviewed staff who had worked in the EPP/EWEP program, most wel-
comed the opportunity to work with welfare recipients as well as
applicants. Additionally, the majority of the Employment Preparation
Divison, EDD and EWEP staff supported the addition of the CRU job functions
to the EPP/EWEP model. They perceived CRU staff's role as consisting of
referring registrants to education and training as well as keeping detailed
records of registrants' activities.
Interviews also indicated that staff were aware of SWIM's 75 percent
participation goal and perceived it to be an achievable one. However,
staff did not pay much heed to it. In the first place, they believed both
that they were working with all registrants referred to their particular
-67- 115
program component and also that they, in turn, referred all those who
completed their component to the next component in the program model.
Staff who had been involved in the EPP/EWEP program indicated that they had
operated in a similar manner prior to SWIM. In the second place, they had
no way of knowing the extent of participation in SWIM.
In fact, CRU staff were the only staff members who could have had a
sense of the extent to which program eligibles were participating in SWIM,
by virtue of their overall activity tracking role. CRU staff did not focus
on this, however, for two reasons.
First, as was clear in interviews, CRU staff did not generally think
in terms of the proportion of their caseload who were participating during
a given month. Like other program staff, CRU staff reported that they were
generally working with all registrants who reached the education and
training stage of the model. (However, in the later months of the deriim-
stration, some CRU staff indicated that their large caseloads were starting
to prevent them from working with all registrants.) Fran the perspective
of the staff, a registrant would be considered active even if he/she was in
the process of being assessed, assigned to a component that was to begin
the following month, awaiting verification of participation, in a
conciliation status or pending deregistration. These registrants would be
considered :worked with."
Second, due to programming problems with the county computer routines
that produced estimates of monthly participation rates, these rates were
communicated infrequently to program staff. Consequently, staff were not
aware of how far they were from the 75 percent demonstration goal on a
regular basis.
Morale among most staff units involved in SKIM was high. Employment
Preparation Division staff, EDD staff and EWEP staff perceived SWIM as an
extension of the EPP/EWEP program which involved little change in their
daily work responsibilities. Additionally, these staff felt that they were
involved in a successful program. Not only did they see registrants
finding jobs, but they also felt that the earlier MDRC evaluation of the
EPP/EWEP program confirmed that, in fact, their portion of the program
'worked.' Lastly, most staff perceived SWIM as the precursor to
California's Greater Avenues for Independence (GAIN) program (see
description in Chapter 1). This added to staff's sense of the importance
of their work.
CRU staff members were the exception; their morale was not as high.
Most of the JDCs had anticipated that their jobs would entail extensive
registrant contact and counseling. In reality, as noted above, program
tracking and monitoring functions consumed the majority of CRU staff time.
One reason for this was that many registrants found jobs or left welfare
for other reasons before reaching the education and training portion of the
program model. Thus, the number of registrants who required assessment and
placement in community programs was limited. Another reason was that CRU
staff spent a good deal of time ensuring that the automated tracking system
contained complete and accurate information and registrants were not
getting lost' in the program. Although the automated SWIM tracking system
was intended to aid in this task, as noted, program resourc were not
available to develop computer routines that would allow the tracking system
to do sane of the clerical work performed by the CRU staff.
-69- -/A 7
CHAPTER 4
THE NATURE OF PROGRAM SERVICES
This chapter describes each type of service provided in SWIM. The
first section describes registration, orientation and appraisal procedures.
The second examines job search activities, regardless of when they occurred
in the model. The third presents details on the EWEP program. The fourth
addresses the nature of education and tra;.ning activities in SWIM. The
final section summarizes the support services available in SWIM.
I. Registration, Orientation and Appraisal
As noted in Chapter 2, the first step in the SWIM program model was
WIN registration. All WIN-mandatory applicants and recipients recently
determined to be WIN-mandatory were informed of their need to register with
the program during interviews with Tncame Maintera (IM) staff.1,2
Recipients who were renewing their previcols WIN/EPP registration were
informed of the need to reregister and given registration appointments by
social workers through the mail. Staff idorttified individuals who were
required to reregister through lists generated by the county's automated
welfare eligibility system.
Program registration was mandatory for Al of these groups. If
applicants did not register, their application for aid could be denied.
Recipients could be sanctioned for not attending their registration
appointment. IM staff were responsible for identifying applicants and
redetermined recipients who failed to register; social work staff were
-70-
18
responsible for identifying renewal recipients who failed to register.
Renewal recipients were given two chances to show up for registration.
After two sequential missed appointments, individuals would be requested to
attend a determination interview ("determining! whether the individual had
a legitimate cause for missing the appointments).3 Generally, if an indi-
vidual attended an appointment at any time during this process, including
the determination interview, the individual would be determined to have a
!good` cause. If an individual never responded to the appointment requests
or never contacted a social worker, a sanction would be requested.
Approximately 25 individuals were scheduled for each registration/
orientation session at each office. 4 Upon arrival for the registration
appointment, the first one-on-one contact with program staff would be with
Coordination and Referral Unit (CRU) staff. 5 Staff would meet for several
minutes with each individual to complete the one-page Client Information
Sheet which elicited demographic information for the research.6
The next step in the process involved completing several forms during
a group orientation session.? Orientation sessions, conducted by social
workers, lasted from 20 to 40 minutes. The purpose of the orientation was
to describe the program, explain the registrants' rights and respons-
ibilities, and provide information on available social services.
Program descriptions given during orientations were brief and low-key,
however, with the emphasis on finding jobs through job search workshops;
little attention was paid to explaining welfare or WIN/SWIM procedures.
The existence of other program components also was not stressed: PUP was
mentioned briefly, if at all, and references were rarely made to possible
education or training opportunities. Thus, registrants were given little
-71- 11D
indication that the program required participation for as long as they
remained on welfare.
Following orientation, all registrants were individually appraised by
social workers.8 For registrants who had been assigned to the control
group, these appraisals were more like exit interviews: Social workers
asked about the families' health, any housing problems, and wheUier there
were any barriers preventing the registrants fran seeking work. Social
workers also reminded registrants that they were entitled to 'entered
employment,' money and supplies if they found jobs. At the conclusion of
the interviews, controls were told that they had met their program
(WIN/SWIM) obligation and were free to leave. There was no further contact
between the controls and program staff unless it was initiated by the
registrant at a later date, for example, to request help with social
services or obtain 'entered employment monies.9
For registrants who were eligible to receive SWIM services, the
appraisal included a determination of their ability to participate in job
search workshop, job club or EWEP; an evaluation of current registrant-
initiated education/training or employment; and the identification of
social service needs. Childcare was also discussed, but not arranged,
during this interview.
At the conclusion of the appraisals, which generally lasted between
five and 20 minutes, SWIM-eligible registrants were assigned to a program
activity, referred to CRU staff, or, very infrequently, deferred fran all
program participation. The guidelines used to make these assignment or
deferral decisions are outlined below.
If, during the course of the interview, social worker: termined that
-72-
120
registrants had no barriers to participation and had not participated in a
job search workshop within the previous year, registrants would be referred
to EDD staff for assignment to a workshop.1° Registrants who
previously participated in a workshop, but not in EWEP, were assigned to
the job club component, along with EWEP. Registrants who had previously
participated in both workshops and EWEP were assigned only to the job club
component.
Registrants who were identified as currently involved in an education
or training program that met the program's deferral standards concerning
type, intensity and duration were referred to CRU staff. Registrants who
were employed at least 15 hours per week were also referred to CRU staff.
Registrants referred to CRU staff were interviewed shortly after their
appraisal. CRU staff requested information fran registrants for use in
verifying their education, training or employment on a periodic basis and
completed a tracking system form to record this self-initiated activity.
At this interview CRU staff usually counseled registrants about better job
opportunities in their field if they were employed, or encouraged
registrants to remain in their education/training program if they thought
it was an appropriate one for the registrant.
If a registrant was identified during appraisal as an undocumented
worker, an individual excluded fran the welfare assistance unit (*excluded
parent'), or a registrant requiring assistance fran a social worker to
remove an impediment to participation, he/she could be deferred fran all
program participation.11 Otherwise, every registran was assigned to sane
program component.
A review of the program case files of a random sample of 242
-73- 1°
individuals who registered with the program between January and March 1986
indicated that only a small group 7 percent were deferred from all
program participation at the conclusion of the appraisal. Among this small
group, over two-fifths were undocumented workers. The only other sizable
group -- 14 percent of those deferred -- had a disability or health
problem. The remaining deferrals reflected a variety of situations.
Individuals assigned to program activities at the conclusion of the
appraisal interview were scheduled for the first available session of the
activity following the ,appraisal. Registrants were generally assigned to
job search workshops which were scheduled to begin approximately three
weeks later. Similar time lags existed between appraisal and initial job
club sessions and MEP orientations.
II. Job Search Activities
Four different types of job search activities were part of the SWIM
program model: job ....Ahops, job clubs, STAR and ISESA. This
section briefly describe; each.
A. Job Search Workshops
Two-week job search workshops were the first program activity for many
registrants. These were held both in English and Spanish. All workshops
used the same materials and essentially the same format, New workshops in
English were begun each week in each oC the two SWIM offices. According to
staff these averaged 10 to 18 registrants. Workshops in Spanish were begun
once a month in each of the two offices. These averaged 20 to 30
regizor-nta.
The first week of es,.: workshop consisted of structured group
-74-
122
instruction, focusing on skills assessment, goal setting and the completion
of employrent applications. Participants were taught in daily sessions how
to write resumes, locate job leads, handle an interview and dae the tele-
phone to obtain appointments.
In the second week, participants were expected to make contacts with
prospective employers. For the English- speaking workshops, participants
worked in a room with a bank of telephones, generally for two hours a day,
placing calls to prospective employers and setting up job interviews.
There was no set quota of calls, but participants were expected to contact
from ten to 20 employers daily. Instead of the usual telephone room work,
Spanish-speaking registrants were required to make three in- person employer
contacts per day. Their limitations in English made blind calls to
prospective employers impractical.
After completing the workshop, individuals who had failed to find
employment were scheduled for job club and/or EWEP orientation. Regis-.
trants were assigned to job club sessions scheduled to begin between two
and four weeks later. The lag between the end of the workshop and EWEP
orientation was generally similar.
B. Job Clubs
Registrants could be assigned to job clubs separately, or in conjunc-
tion with an EWEP worksite assignment. Job clubs, which consisted of
biweekly two-hour joh search sessions, were not part of the EPP/EWEP
program model in San Diego. This component was added to the program model
for several reasons. First, program planners viewed job clubs as a way to
improve the monitoring of participation while registrants were in EWEP.
(Staff could inquire about EWEP participation on a biweekly basis, during
-75- 123
the job club sessions.) Second, job clubs were intended to officially
mandate job search on the one day each week that registrants were not
required to participate in EWEP. Lastly, job clubs were viewed as a way of
maximizing registrants' participation following job search workshops.
Since registrants usually did not begin to participate at EWEP worksites
until four to six weeks after the end of the workshops, job clubs were
viewed as a component which could fill this gap. The job club component
operated from the beginning of SWIM through December 1986.
Between eight and 12 registrants attended each job club session. The
content of the job clubs changed over time in both SWIM offices, eventually
evolving into a combination of motivational exercises, use of EDD computer
listings or JDU job banks for job leads, and repeated telephoning to
prospective employers. Leadership and duration of the job club component
differed between the two SWIM offices. In one office, the leadership of
the job club sessions rotated among all staff and registrants would be
required to attend job club sessions as long as they remained in EWEP,
regardless of how long that was. In the other office, one job club leader
was designated for several months at a time and registrants would be
required to attend job club sessions for a 90-day period.12
Upon completion of the job club component (provided that the EWEP
assignment was also completed), registrants were either scheduled by CRU
staff for an assessment interview at a later date or assessed on the day of
the last job club session.
Almost from the beginning of SWIM, program managers and line staff
were not pleased with the job club component. According to staff,
registrants' job search workshop groups were not kept intact during the job
-76-
124
clubs, group cohesiveness failed to develop in job club sessions, and the
component did not seen to motivate registrants to look for jobs.
Registrants also seemed to staff to be confused when assigned to two
program activities -- job clubs and EWEP -- at the same time. Finally,
only about one-third of the registrants assigned to job clubs attended.
For some registrants, job club sessions interfered with family matters that
registrants needed to take care of during their one day off from EWEP. It
is interesting to note, in light of these staff perceptions of ineffective-
ness, that quite a Lew sanctions were imposed for noncompliance in job
clubs.
The county stopped operating job clubs in December 1986. The
component was replaced by a new activity, Skills Techniques Achievema.it
Reviews (STAR), in January 1987.
C. STAR
/' STAR consisted of intensive supervised job search for a maximum of 13
days between the time a reoistrant completed job search workshop and
attended EWEP orientation. Every other day during this period, registrants
attended group motivational sessions for two to three hours. On the
non-group days, registrants were required to look for work on their own.
Job search workshop groups were kept intact for STAR, and the job search
sessions were led by registrants' previous job search workshop leaders. In
essence, this component stretched the job search workshop component into a
four-week workshop, incorporating more individual job search.
D. Noncompliance in Job Search Workshops, Job Clubs and STAR
Employment Development Department (EDD) staff handled noncompliance in
these three activities, as noted earlier. Staff took action on the follow-
-77-
ing types of situations: failure to attend workshops, job clubs or STAR;
spotty attendance (one or more days missed with no good cause); and
dropping out.
The activity leader of each of these activities was responsible for
monitoring participation among the registrants scheduled for the activity.
For example, the job club leader of a particular session would follow -up on
those who failed to attend that session. Registrants were allowed to be
autanatically rescheduled once if they initially failed to show up at an
activity. After the second failure to attend, registrants were referred to
an EDD determinations specialist and a determination interview was
scheduled.
If a registrant attended the interview and was :determined' to have a
legitimate reason for failing to attend, the registrant was assigned to the
next scheduled session of the activity. If a registrant attended the inter-
view and was found to have no legitimate reason for noncompliance, a con-
ciliation agreement was drawn up between the EDD determination specialist
and the registrant. This agreement, which usually stipulated participation
at the next scheduled session of the activity, represented the registrant's
'last chance.' If the registrant- violated the agreement, the EDD worker
would request that IM impose a sanction. If Lhe registrant adhered to the
agreement, the determination process ended. Later noncompliance would
start the process over again.
For registrants who failed to attend the scheduled determination
interview, EDD staff generally tried to contact them or their IM workers to
see if they had left welfare or were no longer WIN-mandatory. If a
registrant could not be reached and still appeared to be WIN- mandatory, a
-78-
126
!silent! determination would be conducted. In other words, paperwork was
completed indicating that the registrant failed to attend the interview and
was therefore determined to have no legitimate reason for noncompliance.
Following this type of determination, a sanction would be requested.
Throughout the above procedures, EDD staff continued to encourage
registrants to participate, Staff would repeatedly try to contact regis-
trants if they failed to attend a determination interview. Staff were also
quite liberal in their definition of legitimate reasons for noncompliance.
If at all possible, staff tried to conciliate with registrants rather than
request that they be sanctioned.
E. ISESA
Individualized Supervised Employment Search Activity (ISESA) began in
November 1985. This activity, which was a post-assessment option in the
SWIM model, consisted of job search workshops, held one day per week, for a
period of 13 weeks. ISESA was funded through JTPA and operated by the Job
Development Unit (JDU) within the county welfare department. The activity
was designed for SWIM registrants, and almost all participants were
referred by CRU staff.
ISESA was an open-entry, open-exit program; registrants could start
attending sessions at any point during the 13-week curriculum. Registrants
initially attended a short orientation session and then were assigned to a
weekly session. Between ten and 20 registrants attended each session and,
depending on numbers, registrants were occasionally divided into AFDC and
AFDC-U groups. The workshop sessions varied in length, depending on the
material to be covered and registrant interest, but never lasted more than
a few hours.
-79-12 7
As part of each ISESA session, registrants worked with JDU job deve-
lopers to identify possible job openings, either unsubsidized jobs or on-
the-lob-training (OJT) positions.13 Between weekly sessions, registrants
were expected to make ten job applications, some of which were verified by
the group leader.
As a v7nipient of JTPA funds, the JDU was subject to JTPA performance
standards. In order to assist the JDU in meeting their goals, SWIM program
guidelines specified that registrants who had a high school diploma or an
employable skill were to be referred for job search, preferrably through
the JDU but possibly through EDD, following assessment. Sane CRU staff,
however, disregarded this guideline as not being in the best interest of
registrants and referred same of the registrants to education or training
programs.
Participation in ISESA was mandatory for all referred registrants and
was monitored by JDU staff.14 All registrants who completed the 13-week
ISESA curriculum without finding a job were referred back to CRU staff for
further assessment regarding 'next steps.'
III. Work Experience
The Employment Work Experience Program (EWEP) was a 13-week community
work experience canponent in which participants were expected to work up to
four days a week in either the public or the private nonprofit sector. The
number of work hours was calculated by dividing registrants' monthly
welfare grants by prevailing wage rates (based on registrants' education).
Registrants were assigned to work a maximum of 32 hours per week. Hours
were further restricted by the fact that one day a week was reserved for
-80-
128
ongoing individual job search. County welfare department staff developed
the worksites and subsequently monitored EWEP attendance.
Worksite developers did not experience any problems keeping up with
the flow of EWEP referrals, in part because a 'stock' of worksite positions
had been developed through previous county programs. SWIM was the first
program, however, in which worksites were developed that could accomodate
individuals who spoke only Spanish.
EWEP referrals initially participated in group orientation meetings,
followed by individual sessions with staff. In these sessions,
registrants' work histories and possible barriers to working the required
number of monthly hours were discussed. The sessions did not assess social
service needs in any depth, since these presumably had been resolved
earlier. At the conclusion of the session, registrants were assigned to
worksites. Although location was a factor in these assignments, counselors
attempted to find positions that met the registrants' interests and
backgrounds. Interviews with staff indicated that very few registrants
could not be assigned to a site.
EREP staff handled noncompliance in the component by monitoring two
aspects of EWEP: attendance at the initial EWEP orientation and
participation at worksites. Registrants were generally rescheduled for an
orientation if they missed the initially scheduled one. If they missed the
second date, however, the determinations process began.
Attendance at worksites was monitored through time sheets, completed
by worksite supervisors at the end of each month and mailed to EWEP staff.
According to staff, any more than three missed days of a monthly worksite
assignment usually prompted a call or notice to the registrant and a
-81-120
request for a determination/conciliation interview.
During the determination interview, a concilation plan would be
developed. Conciliation procedures, which were added as a result of
legislation which allowed San Diego to continue EWEP from July :985 through
June 1987, gave registrants a second chance to cooperate before the
imposition of a sanction.15
If a registrant did not attend the determination interview or failed
to comply (for no legitimate reason) with the agreed upon conciliation
plan, a sanction would be requested. As described in Chapter 1,
sanctioning rules were less punitive for AFDC-U registrants who were
noncompliant in EREP as compared to those imposed for this group in the
earlier EPP/EWEP program. This was another result of the above-mentioned
state legislative EwEP waiver. The legislation specified that families of
AFDC-U registrants who were sanctioned for noncompliance in connection with
EWEP should not lose their welfare benefits. To comply with this require-
ment, state aid was used to continue benefits for the families of AFDC -L'
registrants sanctioned in connection with EWEP.
For those who completed EWEP and were still not employed, appointments
for assessment by CRU staff were scheduled. At the completion of EWEP,
workfare staff notified CRU staff that a particular registrant was avail-
able for an assessment. CRU staff scheduled assessment appointments with
registrants through mailed notices.
IV. Education and Training Activities
Emphasis on education and training referrals to community organiza-
tions as part of SWIM was a key departure from the county's previous
-82-
13 0
EPP/EWEP program. SWIM planners envisioned that large numbers of regis-
trants would be referred to these types of programs. In reality, this did
not occur. Aided by a healthy local labor market, many registrants became
employed and/or deregistered from the program before reaching the
education/training portion of the model. In addition, sane registrants
enrolled in these types of activities on their awn, thus not requiring
assessments or referrals by SWIM staff.
A. Develo ment of Linke es between SWIM and Communit. Organizations
Since the earlier EPP/EWEP program did not emphasize referrals to
community education and training programs, the development of linkages
between the SWIM offices and education and training providers began during
the SWIM planning stages and continued throughout the two-year period that
SWIM operated. The county started the SWIM planning process with two
established relationships. The Employment Services Bureau was already
receiving JTPA funds to operate the Job Development Unit (JDU) and the
Employment Services Program (ESP). JDU provided on- the -job training and
occupational skills training. ESP trained AFDC recipients for jobs in the
health care field. However, planners anticipated that these relationships
would not be enough because many SWIM registrants would require additional
types of services, namely, English as a Second Language (ESL) courses,
Adult Basic Education (ABE) programs, GED preparation, and other types of
training.
In order to investigate the types of programs provided by community
organizations and ensure access of SWIM registrants to these programs,
representatives from the SWIM program met separately with staff from the
Regional Employment and Training Consortium (RETC, the JTPA administrative
-83-
s.131
agency in San Diego) and the San Diego Community College District (which
includes community college programs and continuing education centers/adult
schools) as part of the SWIM planning process. These meetings primarily
familiarized community organization administrators with the SWIM program;
very few formal agreements between SWIM and the providers were developed.16
Linkages to community education and training providers were also
sought through personal contacts. Many of the CRU staff, who were respon-
sible for education and training referrals, had previous ties to community
organizations. These staff were instrumental, particularly at the
beginning of SWIM, in developing handbooks which described available
community programs, their entry requirements and contact persons in each
program. Over time, however, many staff developed their own contacts with
each provider.
B. Pre-Referral Assessments
Once individuals finished EWEP and/or their required number of job
club sessions, CRU staff scheduled assessment interviews.17 Between two
and ten post-EWBP (or job club) assessments were scheduled per week.
Probably as a result of the lack of education or training information
presented during SWIM orientations, staff indicated that the assess,..ent
interview was often the first time that registrants became aware of the
fact that they could be assigned to an education or training program as
part of the program -- that is, as a condition of receiving welfare. Thus,
time was taken in the assessments to explain this aspect of the program.
Part of the assessment was also counseling intended to motivate the
registrants. Staff noted that registrants who had reached this portion of
the model were those who had failed to find jobs through the job search
-84-
132
workshops, job clubs and EWEP. These registrants generally had the least
work experience and education, and often were the least motivated.
The assessments, which consisted of one long interview or several
interviews over a period of two weeks or more, were intended to determine
registrants' next steps in the program. In these interviews, CRU staff
tried to get registrants to verbalize their vocational interests, by
discussing registrants' work histories, reviewing registrants' prior SWIM
activities, suggesting different vocational opportunities, and completing
questionnaires and forms.
CRU staff would base their referral decisions on several factors.
Skilled registrants, those with a recent work history, or registrants who
were more eager to find a job than enter skills training, were often
referred to the JDU for the ISESA component. Spanish-speaking registrants
were generally referred to English as a Second Language classes, often
concurrently with job training if the registrant could speak some English.
Several CRU staff emphasized ABE and GED classes if th-e-regkstrant-lacked° a.
high school diploma.
To varying degrees, depending on the CRU staff person, registrants
were referred to community college vocational counselors for vocational
assessments.18 Counselor recommendations were given to the registrants to
take to their CRU worker; these counselors did not make direct referrals.
Once a referral was agreed upon between registrant and CRU worker, a
one-page amployment and training plan would be completed and signed by both
the registrant and the CRU worker. CRU workers would then either just send
the registrant to the provider with a referral form, call a personal
contact at the provider agency to inform them of the impending referral, or
133-85-
occasionally 'walk: the registrant over to the provider.
In interviews, staff indicated that they were able to find suitable
programs for all assessed registrants. First, education and training
providers rarely rejected a referred registrant due to unsuitability for a
program. This was at least in part because CRU staff were always careful
to review any eligibility requirements imposed by the provider before
making a referral, since a rejection would start the entire assessment/
referral process over again. Second, providers rarely rejected a referred
registrant due to a lack of program :slots." Community resources were
sufficient to absorb all referred registrants.
Interviews with staff indicated that they scheduled assessments for
all registrants who completed the earlier components. However, many
registrants, at times as many as half, failed to attend their originally
scheduled assessment.19 If the registrant attended any of the several
subsequent appointments scheduled as a result of the initial no-show, the
assessment -interview- would- -be- Aheld- at that time if the CRU worker was
available.
However, according to the CRU staff, rarely were registrants sanction-
ed for failure to attend the assessment: Most registrants eventually
cooperated and therefore staff were hesitant to refer registrants to the
social workers to begin the adjudication process. Although interviews with
CRU staff indicated that they could not easily characterize registrants who
tended to miss assessment appointments, about half of the staff noted that
no-shows tended to be those who had been noncompliant in previous SWIM
components.
-86-
ongoing individual job search. County welfare department staff developed
the worksites and subsequently monitored EWEP attendance.
Worksite developers did not experience any problems keeping up with
the flow of EWEP referrals, in paLt because a ,'stock' of worksite positions
had been developed through previous county programs. SWIM was the first
program, however, in which worksites were developed that could accomodate
individuals who spoke only Spanish.
EWEP referrals initially participated in group orientation meetings,
followed by individual sessions with staff. In these sessions,
registrants' work histories and possible barriers to working the required
number of monthly hours were discussed. The sessions did not assess social
service needs in any depth, since these presumably had been resolved
earlier. At the conclusion of the session, registrants were assigned to
worksites. Although location was a factor in these assignments, counselors
attempted to find positions that met the registrants' interests and
backgrounds. Interviews with staff indicat4td that very few registrants
could not be assigned to a site.
EWEP staff handled noncompliance in the component by monitoring two
aspects of EWEP: attendance at the initial EWEP orientation and
participation at worksites. Registrants were generally rescheduled for an
orientation if they missed the initially scheduled one. If they missed the
second date, however, the determinations process began.
.'ttendance at worksites was monitored through time sheets, completed
by worksite supervisors at the end of each month and mailed to EWEP staff.
According to staff, any more than three missed days of a monthly worksite
assignment usually prompted a call or notice to the registrant and a
1 3 5-81-
request for a determination/conciliation interview.
During the determination interview, a concilation plan would be
developed. Conciliation procedures, which were added as a result of
legislation which allowed Sal Diego to continue EWEP from July 1985 through
June 1987, gave registrants a second chance to cooperate before the
imposition of a sanction.15
If a registrant did not attend the determination interview or failed
to comply (for no legitimate reason) with the agreed upon conciliation
plan, a sanction would be requested. As described in Chapter 1,
sanctioning rules were less punitive for AFDC-U registrants who were
noncompliant in EWEP as compared to those imposed for this group in the
earlier EPP/EWEP program. This was another result of the above-mentioned
state legislative EWEP waiver. The legislation specified that families of
AFDC-U registrants who were sanctioned for noncompliance in connection with
EWEP should not lose their welfare benefits. To comply with this require-
ment, state aiJ was used to continue benefits for the families of AFDC-U
registrants sanctioned in connection with EWEP.
For those who completed EWEP and were still not employed, appointments
for assessment by CRU staff were scheduled. At the completion of EWEP,
workfare staff notified CRU staff that a particular registrant was avail-
able for an assessment. CRU staff scheduled assessment appointments with
registrants through mailed notices.
IV. Education and Training Activities
Emphasis on education and training referrals to community organiza-
tions as part of SWIM was a key departure frcm the county's previous
-82- 1 3
C. Types of Community Programs Utilized
As a result of the assessment interviews, registrants were referred to
either ISESA at the JDU or to community providers of education or training.
The majority of those referred to community providers were referred to
education programs, as opposed to training programs. Fc: both education
and training, programs within the community college systtA were used much
more frequently than JTPA programs. SWIM registrants often did not meet
the entry requirements of JTPA contractors.
Interviews with CRU staff and periodic reviews of registrants' case
files indicated that among those referred to education programs, almost
half were referred to ESL classes; about one-third %ere in GED preparation
courses; and the remaining registrants were referred to ABE programs.
Those referred to training courses were enrolled in a variety of programs,
including nursing aide, accounting, autobody repair, security guard,
electronics, front desk clerk and computer training.
Registrants were referred most commonly for services to branches of
the San Diego community college continuing education system, which
historically has had extensive experience in providing services to
low-income individuals. SWIM registrants were enrolled in the regular
curriculum. In fact, SWIM-referred students were a very small proportion
of all students at these centers.
The continuing education centers offe_ed a variety of programs and
courses: ESL, ABE, GED, and various vocational programs (including
vocational ESL). All programs were open entry an open exit. The hours of
training or instruction varied, depending on the type of program."
Students generally progressed at their an pace; thus length of stay in the
137-87-
program varied by individual.21 The type of instruction -- group sessions
or individual study -- also varied by program. Vocational training
programs were likely to involve a combination of lectures and !laboratory!
work. ABE courses tended to involve individualized instruction. Due to
the wide range of courses offered at the continuing education centers, few
entry requirements existed. If individuals referred to a vocational
training program did not have sufficient language or reading skills, they
were placed in an ESL or ABE course until their skills reacted the required
level.
Starting in January 1987, another referral option became available to
CRU staff. Using JTPA Title II-A 8-percent monies, the county established
three learning center laboratories as a pilot project for GAIN. In total,
the learning centers could serve 100 individuals, divided fairly evenly
among the three sites. SWIM registrants were given priority for the 100
program 'slots.°22
The laboratories operated on an open entry/open exit basis and used a
computer-operated competency-based curriculum. Individuals remained in the
laboratory between two and five hours daily. All students in the program
were WIN-mandatory AFDC or AFDC-U recipients who had been referred to the
program by county welfare staff. The majority of those in the program were
SWIM registrants.
D. Procedures Used to Monitor Attendance
In general, staff monitored attendance, not progress, in education and
training programs. Most CRU staff monitored participation by mailing
school verification forms to referred registrants with instructions to have
school staff verify participation. Registrants would obtain a school staff
signature and return the forms to the CRU worker by mail. 23 These verific-
ation forms were generally issued shortly after the initial referral to
determine whether the registrant had initally enrolled, and every 30 to 45
days after the referral. There were, however, same variations to this
procedure.
At one continuing education branch, the daily attendance of SWIM-
referred individuals, as well as the progress registrants were making, was
monitored by teachers and reported monthly to the CRU worker who served as
the branch liason. In this way, CRU staff were aware of absentee problems
as well as the progress of registrants referred to this branch.
A few CRU staff members occasionally relied on registrants to certify
their participation in education or training programs. Other staff members
noted that they occasionally checked attendance through periodic telephone
calls to the schools. With the exception of the school in which daily
attendance was reported, school staff rarely initiated contact with SWIM
staff concerning attendance problems; school staff would simply note on a
subsequent verification form that the student was no longer in attendance.
Interviews with CRU staff indicated that once a registrant began a
SWIM- ref erred activity, they rarely dropped out without good cause.
However, same staff noted that it frequently took a while for referred
registrants to attend a program. The most common reasons for registrants
dropping out of programs or having prolonged absences included health,
transportation, childcare or family problems.
If a lack of attendance was identified, staff requested that regis-
trants attend meetings in the SWIM office to diucuss their situation. In
these meetings, staff investigated the reason for the participation
-89- 13D
problems and warned registrants that they could be sanctioned for lack of
cooperation. Sane staff would refer to the Employment and Training Plan
signed at the assessment as a 'contract' and tell registrants they had a
contractual obligation to live up to the agreement.
However, failing to attend or dropping out of an education or training
program was not, in and of itself, grounds for sanctioning. CRU staff
could refer registrants to social workers for sanctions only after the
registrant had missed two sequential office appointments. Social workers
then scheduled a reappraisal interview. If two of these interviews were
missed, a sanction could be imposed.24
E. Treatment of Individuals who Completed Education/TrainingPrograms
Generally, if a registrant completed an education or training program
without finding a job, he/she was assigned to -nother SWIM component.
Often CRU staff referred these individuals to job search activities,
provided by a variety of agencies. However, most CRU staff indicated that
registrants often found jobs on their emn once they completed a g;.cgram.
F. Self-Initiated Education and Training
Registrants could enroll in education and train_nq activities, on
their, awn intiative, at any point during their SWIM tenure. If these
activities met program guidelines, participants were deferred frt- other
types of SWIM activities. Education programs were approved (as a deferral
activity) if the registrant was enrolled for at least nine semester units
at an accredited institution of higher education. Training 1.;ograms were
approved if the registrant was in the last semester of training or in the
final four months of training (if the school did not operate on a semester
-90-
140
basis).25
Reviews of registrants' case files indicated that self-initiated
individuals were enrolled in such training programs as word processing,
nursing assistance, refrigerator repair, cosmetology, electronic repair,
and security guard licensing. Among thoea who were enrolled in education
programs, about half were taking college-level courses.
The extent to which SWIM 'encouraged` registrants to enroll in
education or training programs, on their own initiative, cannot be easily
assessed. Interviews with program staff indicated that over the course of
the demonstration, registrants became increasingly aware that self-
initiated activities would defer then from program-arranged components. If
a registrant inquired about education or training opportunities during
his/her post-orientation appraisal, staff explained that referrals to these
types of programs would be made later in the program model but that
registrants could enroll themselves in these types of programs at any time.
Staff also explained, however, that support service monies could not be
paid to registrants who enrolled on their own.
The majority of registrants in self-initiated education or training
were attending programs provided by the community college system; very few
were in JTPA programs. Regardless of the provider, case file reviews
indicdted that the majority of those in self-initiated programs were in
training rather than education programs.
The at lance of individuals in self-initiated programs was
monitored, eN, ,ry 30 to 45 days, through school verification forms mailed to
the students, Students were required to obtain a school staff signature on
the form and return it to the CRU staff. According to staff interviews,
very few self-initiated registrants failed to comply with the verification
procedures.26
V. Support Services
In SWIM, several types of support services were available: transport-
ation monies, incentive payments, !entered employment,' payments and child-
care reimbursements. is illustrated by Table 4.1, however, these allow-
ances and support services were not available to registrants participating
in all SWIM components. This section of the chapter summarizes these
services by type.
A. Transportation
Registrants participating in job search workshops, job clubs, STAR,
EWEP, ISESA and program-referred training were eligible for transportation
payments through EDD and the welfare department. Participants in program-
referred education or self-initiated education/training as well as those
employed while registered were not eligible for these monies.
Registrants participating in job search workshops, job clubs, STAR,
and program-referred training were automatically paid $5 per day of
attendance. Regulations concerning EWEP transportation payments changed
over time. Prior to September 1986, EWEP participants who used public
transportation were paid approximately $2 per day of attendance at a
worksite; those who used their own cars were paid approximately 20 cents
per mile. Starting in September 1986, reimbursement levels remained the
ze but the payment base became the number of days an individual was
assigned to a worksite. Participants in the ISESA component were offered
six bus tokens for each week of the component.
-92-
142
TABLE 4.1
SWIM
SUMMARY OF ALLOWANCES AND SUPPORT SERVICES,
BY TYPE OF SERVICE AND COMPONENT
Component
Type of Support Service
Transportation Incentives
'Entered Employ
ment' Stipends Childcare
Job Search workshop
$5/day of
attendance none available
$5/day, up to
first paycheck
(maximum = $50)
plus needed tools,
unifarms, books $1.25/child/hour
EWEP
$1.60$2.00/day
far bus travel;
$.20$.21/mile for
auto use; paid per
participated day
prior to
September 1?86,
per assigned day
as of September
1986 none available
$5/day, up to
first paycheck
(maximum = $50,
plus needed tools,
uniforms, books $1.25/child/hour
Job Club
$5/c1fly of
attendance none available
$5 /day, up to
first paycheck
(maximum = $:0)
plus needed tools,
unifarms, books $1.25/child/hour
STAR
$5 /day of
attendance none available
$5 /day, up to
first paycheck
(maximum = $50)
plus needed tools,
uniforms, books $1.25/child/hour
ISESA
some bus tokens
given out none available
$5/day, up to
first paycheck
(maximum = $50)
plus needed tools,
uniforms, books $1.25/child/hour
ProgramArranged Education none available none available
$5/day, up to
first paycheck
(maximum = $50)
plus needed tools,
uniforms, books $1.25/child/hour
-93-
(continued)
1 4 J
TABLE 4.1 (continued)
Component
Typo of Support Service
Transportation
'Entered Employ-
Incentives ment* Stipends Childcarea
Program-Arranged
Training
$5/day of
attendance
$1.50/day of
attendance
$5/day, up to
first paycheck
(maximum = $50)
plus needed tools,
uniforms, books $1.25 /child /hour
Self-Initiated Education
or Training none available none available
$5/day, up to
first paycheck
(maximum = $50)
plus needed tools,
uniforms, books none available
Employment While
Registered none available none available
$5/day, up to
first paycheck
(maximum = $50)
plus needed tools,
uniforms, books none available
SOURCE: Program documents and interviews with program staff.
NOTES:aDuring the second year of SWIM, a maximum of $250 per month per child was allowed.
1:494
B. Incentive Payments
Incentive payments of $1.50 per day of attendance were available only
to registrants participating in program-arranged training. As was the case
in the EPP/EWEP program, EDD handled authorization of these payments.
C. 'Entered Employment' Payments
These payments consisted of work expense advances for those who found
jobs. The money was intended to defray work expense costs until the regis-
trant received his/her first paycheck. All registrants -- experimentals as
well as controls -- were eligible for these payments. Once a registrant
found a job, he/she was eligible for a work expense advance of $5 per day
(up to a maximum of $50) for each working day prior to receipt of his/her
first paycheck. Again, EDD authorized these payments.
D. Childcare
ChP,dcare monies were available to registrants in all activities in
the SWIM model, with the exception of self-initiated education/training and
employment while registered. Individuals were eligible for these monies
only while registered with the program; SWIM did not provide any
'transitional" support services. These m '2nies, paid through EDD and the
welfare department, were available for all children under 14 years of age.
Participants accessing EDD child care monies (i.e., those in job
search workshops, job clubs or STAR) could be paid for expenses in advance.
During the first year of SWIM, $1.25 per child per hour could be paid; in
the second year, regulations were changed to allow a maximum of $250 per
month per child. The location of the childcare did Lot affect a
registrants' eligibility for childcare monies; care could be provided in
the child's home or the provider's home.
145-95-
Participants accessing welfare department childcare monies (i.e. those
in ENEP, program- arranged education or training, or ISESA) would ne reim-
bursed for childcare expenditures; payment would not be made in advance. A
maximum of $250 per month per child was allowed. To be eligible for reim-
bursement, the care could not be provided in the child's home.
As much as possible, program staff tried to schedule SWIM activities
around the school hours of registrants' children. (As noted in Chapter 2,
only 10 percent of the registrants in the generally single-parent AFDC
hous.holds had any children younger than six years of age.) Staff reported
that this was possible for most registrants. Rarely did staff ---te that
they could rot "place' or assign a registrant due to a lack ..)f childcare.
However, regardless of whether a child was in school or in a special
childcare arrangement while his/her parent was involved in a SWIM activity,
social work staff frequently monitored the child's situation. Typically, a
social worker would discuss childcare arrangements with registrants during
th.t. post-orientation appraisal interview, on the first day of a job
search workshop, on th, mast day of the workshop, and during each biweekly
job club session (which was concurrent with EWEP, if a registrant had been
assigned to EWEP). So( _a) workers contacted registrants acti to educa-
tion or training (either by telephone or, if the worker suspected problems,
in person) every two months to discuss childcare arrangements. After each
childcare discussion, social workers were supposed to complete a standard
childcare assessment form which would be filed in the registrant's program
case file.
-96-146
CHAPTER 5
OVERVIEW OF REGISTRANT FLOW THROUGH THE PROGRAM
This Chapter traces the experiences of SWIM registrants in the months
following their initial program registration. The focus is on the extent
to which registrants participated in the numerous components of SWIM -- job
search workshops, work experience, job clubs, STAR, program-arranged educa-
tion or training, or other types of job search -- as well as self-initiated
education/training or employment. This analysis identifies typical
patterns of service receipt among program-eligibles and informs the find-
ings on program impacts presented in Chapter 8, by describing the extent to
which individuals in the experimental group received program services. The
results of a special study of the childcare arrangements of SWIM partici-
pants are also presented.
The first section of the chapter provides an explanation of the types
of participation measures presented in this chapter. The second section
presents an overview of the participation patterns of all SWIM-eligible
AFDC and AFDC-U registrants and two important subgroups -- welfare
applicants and recipients. After this broader perspective, the sequence in
which registrants moved through the program model is tracked. This is
followed by a discussion of the duration of services in SWIM -- whether
individuals participated one day, one week or several months. The last
section of the chapter presents the childcare study results.
Chapters 5, 6 and 7 use different types of participation measures in
order to answer different types of research questions. Table 5.1
147-97-
143
TABLE 5.1
SWIM
SUMMARY OF PARTICIPATION MEASURES USEDIN THE SWIM REPORT
Measure Definition Questions Answered Using Measure
Longitudinal ActivityMeasures
These measures focus on a group of Indivi-duals who entered the program during o
specified time period and follow theseindividuals for a certoin number of months(in SW1X. for 12 months) otter programentry.
What percent of registronts 'ever' portl-cipated In the program? In what types ofcomponents were individuals most likelyto participote? To what extent wereindividuals' AFDC grant amounts decreasedbecause they did not cooperate with theprogram? What percent of registrontsleft the program?
Ongoing PorticipatIonMeasures
Monthly ParticipotionMeasures
These measures focus on o group of indi-viduals who entered the program during a
specified time period and follow theseindividuals for o certoin number of months(in SWIM, for 12 months) after programentry. ?toy examine the proportion ofprogrom-eligible months in which Indivi-duals porticipated.
To what extent did individuals portici-pate during every month they wereeligible for the progrom?
These meosures focus on the group ofindividuals eligible for the progromduring o given month ond examine the
proportion who were active during thatmonth.
To what extent did the provom soturotethe WIN - mandatory coselood with employ-
mont-enhonclng octivitiv, during o givenmonth? In any month, what types ofprogram services were utilized? How didthe arroy of rendered progrom serviceschonge over time?
149
summarizes the various measures. In this chapter, longitudinal activity
measures and measures of the duration of services are used. These measures
indicate the percentage of registrants who 'ever' (that is, within 12
months of initial registration) participated in the program, the types of
components in which individuals were most likely to participate, the order
in which registrants typically proceeded through the program model, and the
average length of time registrants ever participated in the program.
Chapter 6 addresses the extent to which the program implemented a
continuous participation requirement. To measure this, the proportion of
program-eligible months in which individuals participated is examined.
Chapter / then analyzes the extent to which SWIM 6staturatedu the WIN-
mandatory caselc with employment-enhancing activities. To do this,
monthly participation measures are used. These represent the proportion of
individuals eligible for the program during a month who were active in the
program during that month.
I. Explanation of Longitudinal Activity Indicators
Longitudinal activity measures focus on a group of individuals who
enter a program during a specified time period and follow these individuals
for a certain number of months after program entry. It is useful to begin
the discussion by illustrating has the 12-month participation rates
present in this Chapter are calculated. Figure 5.1 depicts the program
experiences of ten hypothetical SWIM registrants. These ten individuals
registered at different points in the SWIM program and had varying
participation patterns after registration. In order to examine the
participation patterns of all registrants for a full 12 months following
-99-150
Person A
Persoo B
Person C
Person D
Person E
Person F
Person G
Person H
Person I
Person J
FIGURE 5.1
SWIM
ILLUSTRATION OF TIME PERIODS COVERED BY TWELVE-MONTH LONGITUDINALACTIVITY MEASURES USING TEN HYPOTHETICAL SAMPLE MEMBERS
1985 1986Jul Sep Nov Jan Mar May Jul Sep Nov Jan Mar May
, I I , t
1987
1;.,...,.:. '
: : : :
. . . .....F-H
H-1----1
Registered andParticipating
Registered andNot Participating
Time Period Usedin Calculating Rates
152
initial registration, only individuals who registered with SWIM by June
1986 can be counted. Persons G, H, I and J, who registered after this
date, are excluded fran the calculation of 12-month participation rates.
This type of measure takes into account the activities of each registrant
during the 12 months following registration. This time period, for each
registrant counted in the rate, is shown by the shaded area in the figure.
Thus, for the examples shown, five out of six individuals (or 83 percent)
participated at sane point within 12 months of registration.
These are the types of participation rates presented in the first
section of this chapteq, using the full samlit.e of registrants. As in the
above illustration, the focus is on those individuals who *egistered during
the first year of the demonstration, the same sample used to calculate the
program impacts presented in Chapter 8. To allow follow-up of more than 12
months, rates are also calculated for an early group of registrants, that
is, those who *-egistered between July 1985 and December 1985.
This simple participation rate, as can be seen fran Figure 5.1, does
not take into account the duration of services -- whether an individual
participated one day, one week, one month or several months. A later
section of this chapter addresses the issue of duration, using the same
samples and follcw-up periods.
II. Participation: An Overall Perspective
Table 5.2 presents 12-month activity indicators, using data fran the
SWIM aLtanated tracking system. 1The broadest indicator of registrant
activity, shown at the top of the table, indicates the proportion of
individuals who, within 12 months after registration, took part in some
-101-153
TABLE S.2
SWIM
TWELVE-MONTH ACTIVITY MEASURES FOR SWIM-ELIGIBLES,lY ASSISTANCE CATEGORY
Activity Measures AFDC AFDC-U
Participated in Any Component, IncludingEmployment While Registered°
Participated in Any Component, ExcludingEmployment While Registered
77.3%
64.4
74.1%
64.5
Participated in Job Sear0 Activities 50.6 56.5***Job Seorch Workshop 41.5 49.0***Job Club 29.6 29.8STAR 0.7 1.0ISESA 5.2 7.4**Union Job Search 0.0 1.3***
Other Job Search 1.2 1.6
Participated in Work Experience 19.5 '19.3
EWEP 19.1 19.0On-the-Job Training 0.7 0.7
Participated in Education or Training 24.3 16.6***Program-Arranged Education or Training 14.3 9.8***Program-Arranged Education 9.0 6.1 **
Provided by Community Colleges 7.3 4.8**Provided by JTPA 0.9 0.7Other Providers 1.4 0.7
Program-Arranged Training 6.1 4.0**Provided by Community Colleges 2.7 2.0Provided by JTPA 1.0 1.3Other Providers 2.5 0.9**
Self-Initiated Educotion or Training 12.8 8.4***Provided by Community Colleges 8.S 6.3*Provided by JTPA 0.4 0.6Other Providers 4.8 2.0***
Employed While Registered° 39.0 34.4**
Moved Out of the SWIM Area 8.0 8.9
Deregistered 61.5 66.3**Due to Sanctioning 10.6 8.4
Sample Size 1608 704
-102- .15 4
(continued)
TABLE 5.2 (continued)
SOURCE: MDRC calculations from the County of San Diego Deportment ofSocial Services SWIM Automated Tracking System and EWE? attendance logs.
NOTES: The sample for this table consists of SWIM-Eligibles whoregistered between July 1985 and June 1986.
Activity measures are calculated as a percentage of the totalnumber of persons in the indicated assistance category. The twelve-monthfollow-up period begins at the point of initial registration.
Participation is defined as attending EWEP for at least one houror any other activity for at least one day.
A chi-square test was applied to differences between assistancecategories. Statistical significance le.els are indicated as: * = 10 percent;** = 5 percent; *** = 1 percent.
aProgram employment information is based on employment that was
reported to program staff. Program employment data were not used to measureimpacts.
155-103-
type of activity while registered with the program. According to this
definition, about three-quarters of SWIM-eligibles participated in job
search workshops, EWEP, OJT, job clubs, STAR, ISESA, other types of job
search, education or training, or employment that did not result in
immediate deregistration 1..,:r^ the program. This proportion was slightly
higher for the AF,,C's than the A2DC-U's -- 77 percent versus 74 percent.
Most of this differeattal is due to the fact that a higher proportion of
AFDC's were employed while registered with the program. This probably
occurred due to the .11110 hours' rule in effect for AFDC-U's. According to
this rule, AFDC-U's who work more than 100 hours in a month became
ineligible for welfare; the rule does not apply to AFDC's.
Evident from this participation rate is that approximately one-quarter
of the sample never participated in any type of activity -- program-
arranged, self-initiated or employment activity -- while registered with
the program. This lack of participation could reflect several types of
situations. First, many of these registrants may have found jobs which
resulted in deregistration from the program before they participated in any
activities. Second, sane may have been denied AFDC or deregistered from
the program for reasons other than employment before participating in any
activities. Third, a small number may have been judged inappropriate for
assignment to a program activity, for example, because they were
undocumented workers. Finally, sane may never have attended the program
activity to which they were assigned and may have been subsequently
sanctioned. These situations are discussed in more detail in Chapter 6.
The second overall indicator of registrant activity shown in Table 5.2
represents a more limited definition of participation. This rate, unlike
-104-
the first rate, does not count employment while registered with the program
as an activity. It Shows that almost two-thirds of the SWIM-eligibles
participated in sane type of job search, work experience, education or
training activity -- a proportion that was almost identical for AFDC's and
AFDC-U's.
Slightly over half of the sample -- 51 percent of the AFDC registrants
and 57 percent of the AFDC-U registrants -- participated in sane type of
job search activity during the one-year follow-up period. As would be
expected in a sequential program model, participation rates were highest in
the first component in the model, that is, job search workshops. (Many
registrants were likely to leave the program before reaching the second or
third components in the model.) Forty-two percent of the AFDC's and 49
percent of the AFDC-U's participated in a job search workshop. 2
Close to 30 percent of the sample participated in job clubs. Partici-
pation in STAR, the component which replaced the job club component in
January 1987, was low. This is primarily because most a the SWIM-
eligibles who registered with the program prior to July 1986 were beyond
the job club/STAR point in the model sequence or had left the program by
January 1987. Participation in job search activities other than job search
workshops, job 'ubs or STAR was not very prevalent among SWIM-eligibles.
Most individuals who participated in other types of job search activities
were active in the 90-day ISESA program. Five percent of the AFDC's and 7
percent of the AFDC-U's participated in this program.
Approximately 19 percent of both AFDC's and AFDC-U's participated in
work experience during the follow-up period, almost all through the EWEP
program. Only a few individuals participated in on-the -job training, which
-105- 157
was provided through the Job Development Unit.
Overall, 24 percent of the AFDC's and 17 percent of the AFDC-U's
participated in education or training activities, either program-arranged
or on the registrant's own initiative.
Slightly over half of all sample members who participated i,s education
or training activities had been placed in these activities by program
staff. Most common was participation in educational programs, which
included Adult Basic Education classes, English as a Second Language
classes, and courses in preparation for General Educational Development
examinations. Nine percent of the AFDC's and 6 percent of the AFDC-U's
attended such courses, most commonly through the community college system.
Six ercent of the AFDC's and 4 percent of the AFDC-U's participated in
E4..,gra1 -arranged training.
Thirteen percent of the AFDC's and 8 percent of the AFDC-U's
participated in self-init_ated education or training activities that were
both known to and approved y the program staff.3
According to program records, over one-third of the SWIM-eligibles
were employed while still registered with the program. Part-time work,
defined as 15-30 hours per week, did not lead to deregistration from the
program. According to the program tracking system data, however, some of
these individuals were employed full-time (more than 30 hours per week).
These individuals probably remained registered with the program only until
IM staff verified their employment and issued a deregistration and/or case
closure.
Seventy percent of the AFDC's and 75 percent of the AFDC-U's became
ineligible for SWIM services at some point during the 12-month follow-up,
-106-
because they either moved out of a SWIM area or were deregistered fran the
program. The majority of those who exited the program did so within six
months of registration; after the sixth month, the rate at which
individuals deregistered fran the program declined.
Over half of all SWIM-eligibles -- 51 percent of all AFDC's and 58
percent of all AFDC-U's -- were deregistered for a reason other than a
sanction: Sane of these deregistrations occurred when people left welfare
entirely, either because they found employment or for other reasons; other
deregistrations took place when clients remained on welfare but were no
longer WIN-mandatory, e.g., because of the birth of a child.
Approximately 8 percent of all SWIM-eligibles moved into the jurisdic-
tion of a non-SWIM program office within the county. Activity data for
individuals served by non-SWIM offices were not collected as part of the
evaluation, but these offices offered many of the same activities a,railable
through SWIM: job search workshops, job clubs, STAR, and EWEP.
Eleven percent of AFDC's and 8 percent of AFDC-U's were sanctioned for
noncompliance with program requirements.4 This proportion is higher than
in most other states studied as part of the MDRC Work/Welfare
Demonstration. 5
Almost all individuals who eventually participated in SWIM did so
within 12 months of registration; very few individuals began participating
after the 12-month follow-up period had elapsed. (See Appendix Table C.1
for 12- and 18-month performance indicators for an early sample of regis-
trants.) It is also interesting to note, however, that sanctioning rates
increased very little with longer follow-up. As discussed in Chapter 4,
this was probably because sanctions were rarely imposed for individuals
-107- 15a
noncampliant in education and training activities.
Recipients were more likely than applicants to participate within 12
months of registration.6 (See Table 5.3.) This pattern was true for both
AFDC's and AFDC-U's. Almost 57 percent of the AFDC applicants participated
in job search, work experience, education or training activities, compared
to 70 percent of the AFDC recipients; the analogous rates for the AFDC-U's
were 63 percent and 67 percent, respectively. A similar pattern was
evident for almost all types of activities within the SWIM model.
The likely explanation for these differences is a rather simple one.
As shown in Table 5.3, applicants were much more likely than recipients to
deregister from the program -- reflecting their higher possibility of leav-
ing welfare and the fact that about 15 percent of the applicants were
denied welfare in the first place. 7 ConsPluently, although both applicants
and recipients were assignru to program activities, recipients were more
apt to participate in them before deregistering.
However, sanctioning rates were higher for applicants than for
recipients, among both AFDC's and AFDC-U's. This may reflect greater
noncompliance on the part of applicants.
III. Sequences of Activities in SWIM
The discussion in the previous section indicated that although almost
three - quarters of the sample members participated in sane type of activity
within 12 months of registration, about one-third of these participants
were never active job search workshops, the expected first component for
the majority of SWIM participants. It is also true that a substantial
proportion 6f the individuals who did attend job search workshops did not
-108- 1"
TABLE 5.3
SWIM
TWELVE-MONTH ACTIVITY MEASURES FOR SWIM-ELIGIBLES,
BY ASSISTANCE CATEGORY AND WELFARE STATUS
Activity Measures
AFDC AFDC -U
Applicants Recipients Total Applicants Recipients Total
Participated in Any Component, including
Employment While Registered 71.7% 81.1% 77.3%*** 71.1% 78.3% 74.1%**
Participated in Any Component, Excluding
Employment While Registered 56.7 69.6 64.4*** 62.6 67.1 64.5
Participated in ...)b Search Activities 45.7 53.8 50.6*** 56.7 56.3 56.5Job Search Workshop 38.8 43.4 41.5* 50.4 47.1 49.0CD
mD Job Club 22.3 34.5 29.6*** 25.4 36.9 29.8***1
STAR 0.9 0.6 0.7 0.5 1.7 1.0ISESA 2.8 6.8 5.2*** 6.1 9.2 7.4Union Job Search 0.0 0.0 0.0 1.7 0.7 1.3Other Job Search 1.2 1.2 1.2 2.2 0.7 1.6
Participated ih Work Experience 15.3 22.3 19.5*** 17.1 22.4 19.3*EWEP 14.8 22.0 )9.1*** 16.9 22.0 19.0On-the-Job Training 0.8 0.6 0.7 0.7 0.7
b0.7
Participated in Education or Training 18.1 28.4 24.3*** 12.7 22.0 16.6***Program-Arranged Education or Training 9.7 17.4 14.3*** 7.8 12.5 9.8*Program-Arranged Education 5.7 11.2 9.0*** 4.9 7.8 6.1Provided by Community Colleges 4.8 9.1 7.3**4. 3.9 6.1 4.Provided by JTPA 0.2 1.4 0.9** 0.5 1.0 0.
7bOther Providers 1.1 1.7 1.4 0.7 0.7 0.7
161
(continued)
162
TABLE 5.3 (continued)
Actigity Measures
AFDC AFDC-U
Applicants Recipients Total Applicants Recipients Total
Program-Arranged Training 4.6 7.1 6.1 3.2 5.1 4.0Provided by Community Colleges 2.5 2.9 2.7 1.2 3.1 2.0Provided by JTPA 1.1 0.9 . 1.0 1.5 1.0 1.3Other Providers 1.2 3.3 2.5 0.7 1.0 0.9
b
Self-initiated Education or Training 9.1 15.3 12.8 6.1 11.5 8.4**Provided by Community Colleges 5.9 10.3 4.2 9.2 6.3Provided by JTPA 0.8 0.2 0.4b 1.0 0.0 0.6
b
Other ProvidorS 3.1 5.9 4.8.* 1.5 2.7 2.0
Employed While Registered° 37.7 39.9 39.0 31.3 38.6 34.4
Moved Out of the SWIM Area 10.8 6.0 8.0*** 10.5 6.8 8.9
Deregi stored 70.0 55.8 61.5 72.6 57.6 66.3**olDue to Sanction 12.8 9.1 10.6 9.3 7.1 8.4
Sample Sl?.e 647 961 1608 409 295 704
SOURCE: See Table 5.2.
NOTES: The sample for this table includes SWIM-Eligibles who registered between July 1985 and June 1986.
Activity measures are calculated as a percentage of the total number of persons In the Indicated welfare status withineach assistance category. The twelve-month follow-up period begins at the point of initial rerlrotion.
Participation Is defined as attending ENEP for at least one hour or any other activity for at least one day.
A chi-square test was applied to differences between welfare statuses within each assistance category. Statisticalsignificance levels are indicated as: = 10 percent; 0 = 5 percent; 00 = 1 percent.
.
(3 -.'U not used to me:=11114:mcPtis
employment Information Is based on ampler/tont that was reported to program staff. Program ameloyment data were
bChl-square test Inappropriate due to lam expected cell frequencies.
participate any further in SWIM. Many of these individuals became employed
and/or left the welfare rolls, thus interrupting their program participa-
tion. This section presents the typical paths that registrants followed
through the SWIM program, highlighting the effect that employment, parti-
cularly part-time employment, had on registrants' experiences in the
program.
The SWIM program model specifies sequences of components, depending on
registrants' activities as of and prior to registration. Program planners
envisioned the following major paths through the SWIM model.
The primary planned SWIM activity sequence, as noted earlier, consist-
ed of job search workshops followed, for those who failed to find jobs
during workshops, by EWEP concurrent with job clubs. For those who
completed EWEP and job clubs without finding employment, referrals to ISESA
and/or education or training programs were to be made. Planners envisioned
that most registrants would follow this activity sequence.
Planners also expected a sizable proportion of registrants to be
deferred from the job search workshops, and possibly from EWEP, because
they had already participated in these activities within 12 months prior to
registration. These registrants were to be assigned to job clubs (perhaps
along with EWEP), followed by referral to ISESA and/or education or train-
ing programs. Other anticipated exceptions to the typical sequence includ-
ed relatively small numbers of registrants who were employed as of registra-
tion, and small numbers participating in self-initiaLea education or train-
ing activitiev as of registration. Once this participation ended, these
individuals were to be assigned to the primary SWIM sequence: job search
workshops; EWEP along with job clubs; ISESA and/or education or training.
Table 5.4 shows the actual. sequences of activities, within 12 months
of registration, for individual: who registered during the first year of
SWIM.8 Note that this table records employment that occurred while an
individual was registered (generally part-time wot...) as well as employment
which had the result of deregistering an individual from the program
(generally full-time work).
As planned, the most common first assignment for registrants was to a
job search workshop. For 36 percent of the AFDC's and 43 percent of the
AFDC -U's, the first activity in which they participated was a workshop.
The next step for those who initially participated in a workshop varied.
Over one-quarter found employment that was known to program staff, either
full-time or part-time, during the course of the workshop or before
participating in another component. Approximately one-third proceeded to
participate in DWEP.
Surprisingly, from the perspective of the program planners, employment
while registered (either part-time or full-time) was the seccnd most common
initial activity. Seventeen percent of AFDC's and 13 percent of AFDC-U's
fit this pattern. These individuals rarely participated in any other type
of activity.
Seven percent of AFDC's and 4 percent of AFDC-U's were participating
in self-initiated education or training as of registration.9 Most
commonly, these individuals remained in education and training throughout
their time in the program; rarely did they participate in other SWIM
activities.
A small number of individuals (3 percent of AFDC's and 2 percent of
AFDC -U's) skipped the job search workshop portion of the SWIM model and
-112 -
16;
TABLE 5.4
SWIM
PERCENTAGE OISTRIBUTION OF SWIM-ELIGIBLESBY ASSISTANCE CATEGORY ANO PARTICIPATION IN A FIRST OR SECOND ACTIVITY
WITHIN TWELVE MONTHS FOLLOWING INITIAL REGISTRATION
First Activity Second Activity I A-DC AFDC -U
Job Search Workshop 35.9 42.8***
EWEP 12.7 13.8
ISESA 0.0 0.0a
Program-Arranged Educationor Training 1.2 0.4
Self - Initiated Education or
Training 1.0 1.7
Employment While Registered 9.7 10.5
Oeregistered Oue to Employment 0.5 1.6*4
No Other Activity 10.7 14.8***
EWEP 3.1 1.7*
aJob Search Workshop 0.1 0.1
ISESA 0.0 0.0a
Program- Arranged Education orTraining 0.4 0.1
Self-Initiated Education orTraining 0.1 0.0
a
Employment While Registered 0.9 0.7
Oeregistered Oue to Employment 0.0 0.0a
No Other Activity 1.6 0.7
Program-Arranged Education
or Training 6.3 4.4*
lob Search Workshop 0.5 0.4
EWEP 0.0 0.0a
ISESA 0.0 0.0a
Self-Initiated Education orTraining 1.5 0.9
Employment While Registered 1.3 0.9
Oeregistered Oue to Employment 0.1 0.0a
No Other Activity 3.0 2.3
1 6 7-113-
continued)
TA8'..E 5.4 (continued)
First Activity Second Activity AFDC AFDC-U
Self-Initiated Educationor Training 7.3 4.1***
Job Search Workshop 0.6 0.4
EWEP 0.0 0.0a
ISESA 0.0 0.0a
Program-Arranged Educationor Training 0.4 0.3
Employment While Registered 1.4 1.3
Doregistered Due to Employment 0.0 0.1a
No Other Activity 4.r 2.0***
Employment While Registered 16.9 13.1**
Job Search Worksflop 1.0 1.3
EWEP 0.1 0.1a
ISESA 0.0 0.0c
Progrnm-Arranged Education orTraining 0.8 0.
Self-Initiated Education orTraining 0.5 0.1
DeregistereC Due to Employment 2.8 4.3*No Other Activity 11.8 7.0***
Deregistered Due toEmployment 1.3 3.0***
SWIM Component 0.2 07Employment While Registered 0.5 11..6
No Other Activity 0.6 1.3**
Other Activity Any Activity ;..1.1 3.7
Never Active 24.0 26.6
Teal 100.0 100.0
Sample Size 1508 704
SOURCE: MDRC calculations from the County of Son Diego Department of SocialServices SWIM Automated Tracking System and EWEP attendance logs.
NOTES: The sample for this table includes ZWIA-Eligibles who registeredbetween July 1985 and June 1986.
Distributions may not add to 100.0 percent and 'Second Activity'percentages may not add to 'First Activity' percentages due to rounding.
(continued)
TABLE 5.4 (continued)
Participation in a first or second activity fol!owing initio'registration is defined as attending EWEP for "it least one hour or any other activityfor at least one day.
Program employment information is based on employrnt that was reportedto program staff. Program employment data were not used to measure impacts.
A chi-square test was applied to differences between assistancecategories. Statistical significance levels are indicated as: 41 = 10 percent; = 5percent; *a = 1 percent.
aChi-square test inappropriate due to low expected cell frequencies.
16D-115-
participated initially in EWEP. As discussed earlier, it is likely that
these individuals had already participated in job search workshops within
12 months prior to registration.
IV. Duration of Participation
So far in this chapter, an individual has been considered a partici-
pant if he/she attended an activity for at least one day during the
12-month follow-up period. No distinction has been made between regis-
trants who participated one day and those who participated for 12 months.
This section presents findings on the duration of participation.
This discussion does not take into account the number of months in
which a registrant was el'. Bible for the program. It simply characterizes
the duration of participation within the 12-month follow-up period, since
impacts are calculated for all individuals in the sample, regardless of how
long they remained eligible for the program. (Chapter 6 addresses the
extent to which individuals were active during every month they were
SWIMHeligible.)
Averaging the number of months in which individuals took part in same
type of activity while registered with the program provides an overall indi-
cator of the length of time registrants participated. Included is partici-
pation in job search workshops, MEP, OJT, job clubs, STAR, ISESA, other
types of job search, education or training, or employment while registered
with the program. Using this type of measure, sample members participated
on average during 4.4 months of the 12-month follow-up period. This
statistic includes the one-quarter of the sample who never participated.
Among participants, the average number of ,'active' months was 5.8.
-116- 17)
Excluding months in which registrants' only activity was employment,
sample members participated on average during 3.0 months of the 12-month
follow-up period. This statistic, however, includes the one-thl-i of the
sample who never par.-;cipated in job search, work experience or education/
training. Among participants in such activities, the average number of
:active' months was io5.
The above figures make no distinction between registrants who partici-
pated one day during the month and those who participated every day of the
month. Duration of participation is more evident when each SWIM program
component is considered separately. Each component had different partici-
pation requirements. For example, individuals wh only participated in a
job search workshop, a two-week activity, could not have participated for
more than 10 days during the month. Registrants who participated only in
job clubs, the biweekly job search activity, are likely to have partici-
pated only one or two days during each month. And, registrants in school
may have been enrolled in ESL classes which were held two mornings per week
-- for a maximum of about eight days of the month. In addition, for same
components attendance data were available, but for 01.41ers only enrollment
data were available.
Job search activities were generally the least time-intensive compon-
ents in the model. Among those who participated in job search workshops
during the 12-month follow-up period, over three-quarters remained is the
workshop for the entire ten-day period. On average, job club participants
attended 4.1 job club sessions within the follow-up period.
The extent of participation in ISESA -- which consisted of two- or
three-hour weekly job search sessions along with individual job search
-117-171
between sessions over a 90-day period .- varied among participants. Data
on the actual number of sessions attended are not available. But the
length of tire individuals remained enrolled in this program provides sane
clues as to hc. many sessions registrants were likely to have attended.
About one-third of the participants remained enrolled with the program for
30 days; another 22 percent were enrolled 31-60 days; and the remaining
participants were enrolled for the full duration of the component.10
Among those who participated in EWEP, participation was quite
intensive. The average number of hours worked at a worksite during the
12-month follow-up period was 167: 161 hours for AFDC registrants and 177
hours for AFDC -U registrants. EWEP participants were generally assigned to
work 20-30 hours per week; registrants did not work full-time hours.
Consequently, the average of 168 hours cannot be directly translated into
actual full-time days. However, these hours would be the rough equivalent
of 24 full days of work.
Individuals who participated in education or training programs
generally remained active over a long period of time although participation
was not necessarily full-time. On average, registrants remained enrolled
with community college programs for a period of 195 days within the
12 -month follow -up period; the enrollment period in JTPA programs averaged
88 days during the follow-up period; and enrollment in other types of
programs averaged 138 days within this period. In terms of months,
registrants were active in education or training programs during an average
of 5.5 months within the 12 -month follow-up period.
The length of time registrants remained employed while registered also
varied. However, on average, part-time employment spanned a period of 149
days while full-time snployment covered a period of 94 days. In terms of
months, registrants were employed (while registered) during an average of
4.7 months within the 12-month follow-up period.
V. 1pecial Study of Childcare
Support services are key .oimponents in programs that seek to maximize
participation. The lack of such services is often viewed as the primary
impediment to mandatory programs, particularly those which mandate
continuous participation. This section presents the results of a special
study of one important type of support service in SWIM, childcare.
The purpose of the study was to describe participants' childcare
arrangements. The study did not investigate why participants chose certain
types of care or whether participants were satisfied with their
arrangements. Evidence presented in Chapter 7, however, indicates that
childcare problems were a barrier to participation for only a small number
of registrants.
To investigate the types of childcare arrangements registrants util-
ized in MDRC staff reviewed the program case files of a random sample
of 121 AFDC registrants -- 61 who participated in July 1986 and 60 who were
active in November 1986. 11AFDC-U registrants, who by definition are in
two-parent households, were not included in the sample. Participants were
defined as those active in any type of SWIM component: job search work-
shc,s, job clubs, EWEP, ISESA, program-arranged education /training, self-
initiated education/training, and employment while registered. The case
files were reviewed to ascertain the nature of childcare arrangements
'nring November 1986 (when children were likely to be in school) and during
-119-
1.73
July 1986 (generally a non-school month). On the basis of case notes,
standard childcare assessment forms, and reimbursement vouchers found in
the case files, information was recorded for each child of the
registrant.12 As much possible, data collectors tried to isolate
childcare arrangements for the two months in question.
In terms of childcare requirements, registrants who participated
during the selected months were a diverse group. Among the 121 registrants
in the sample, 50 percent had one child; 31 percent had two children; and
19 percent of the registrants had three or more children. In 7 percent of
the cases, the youngest child of the registrant was under six years of age.
These registrants were probably 'soft mandatories,' i.e., individuals who
were WIN-mandatory because they were in school. For 46 percent of the
registrants, their youngest child was between six and nine years of age.
The youngest child of 26 percent of the registrants was between 10 and 13
years old. For 22 percent of the registrants, all children were at least
14 years old. Overall, data were collected on 213 children. 13
Registrants in the sample also participated in a wide range of activi-
ties during the review months. Forty percent were employed; 26 percent
were in self-initiated education or training; 21 percent participated in
job clubs during the month; 10 percent were in job search worksh)pst 9
percent were active in EWFP; 7 percent participated in program- arranged
education or training; 4 percent had attended an EWEP orientation; and 3
percent were active in other types of activities. Note that many individ-
uals participated in more than one activity during the review months.
The results of the case file reviews, combined for the July and
November samples, are shown in Table 5.5. The childcare arrangements for
-120- 1
TABLE 5.5
SWIM
PERCENTAGE DISTRIBUTION OF CHILDCARE ARRANGEMENTS FOR AFDC PARTICIPANTS,BY PROVIDER AND LOCATION
Provider and Location
ArraNgementsfor
Participant'sYoungest Child
Arrongementsfor
All Children
of Participant
Relative Provided CareIn Child's HomeIn Provider's Home
Location of Care Not round in Casefile
Non-Relative Provided CareIn Child's HomeIn Provider's HomeLocation of Care Not Found in Casefile
Group Care'n Provider's HomeIn a Center
SWIM Activity Usually Occurred While Childwas in His/Her Regular School
Other Core Provided by o RelativeOther Care Provided by a Non-RelativeAdditional Care Provider Not Found in
Casefile
SWIM Activity Always Occurred While Childwas in His/Her Regular School
Child C.J/ed for Self (1/ years or older)
Provider Information for July orNovember 1986 Not Found in Casefile
Total
Sample Size('
28.1
9.1
12.4
6.6
14.01.78.34.1
4.1
2.5
1.7
11.6
0.0
3.3
8.3
6.6
23.1
12.4
100.0
121
30.5
12.7
11.3
6.6
12.7
0.9
7.0
4.7
4.2
2.8
1.4
8.5
0.0
2.3
6.1
5.6
28.6
9.9
100.0
213
SOURCE: MDRC calculations from casefile reviews of randomly chosen .FDCregistrants who were active during July or November, 1986.
NOTES: Distributions may not odd to 100.0 percent and subcategories may notadd to category totals due to rounding.
GFour participants (representing 7 children) who were active during
both July and November 1986 wore randomly chosen for both the July and Novembersamples. Therefore, 4 participants and 7 children are represented twice In thesample size.
175-121-
participants are best characterized by focusing on the situation of the
registrants' youngest child. The results indicate that 30 percent of the
participants did not need childcare during the review months: For 23
percent of the participants, their youngest child was at least 14 years
old.14 For 7 percent of the participants, all SWIM activity took place
while their youngest child and, presumably, all their children, were in
school.
Twelve percent of the participants were active in SWIM while theil
youngest child was in school, but required pre-school, after-school or
'back-up' care. For most of these participants, case file reviews did not
yield further information on the provider.
Over two-fifths of all participants (42 netcent) used informal day
care arrangements for their youngest child. For over two-thirds of those
with informal arrangements (28 percent of the entire childcare sample),
care was provided by relatives as opposed to non-relatives. Among those
whose children were cared for by relatives, the care took place in the
child's home almost as frequently as in the relative's home. Among those
whose children were cared for by a non-relative, the care most commonly
took place in the non-relative's home.
Only 4 percent of the participants placed their youngest child in a
formal group care arrangement. These participants, who generally had
children under six years of age, used facilities at the University of
California - San Diego and the YMCA.
Similar childcare patterns are evident when all children of the
registrants, not just the youngest child, are included in the statistics
(Table 5.5). In fact, 84 percent of the sampled participants had the same
-122- 176
type of childcare arrangement for all their children. The remaining 16
percent of the registrants used childcare arrangements that differed within
the family according to the childrens' ages.
Childcare arrangements differed between 'school' and ,'non- school'
months as shown in Table 5.6, For one-third of the November 1986 partici-
pants, all or part of their SWIM activity occurred while their youngest
child was in school. During July 1986, when children were generally not in
school, informal care provided by relatives or non-relatives became more
common. In July, 52 percent of the participants used informal day care.
In November, 32 percent of the participants used this type of care.
Childcare arrangements also differed alcording to the age of each
child (Table 5.7). The few young children in the sample (less than six
years old) were likely to be cared for in a formal group care arrangement.
These in the six- to thirteen-year-old age group were most likely to be in
school and/or cared for by a relative.
No clear relationship was evident between childcare arrangements and
the type of SWIM activity in which the registrant was participating during
the review month. The c,%a exception was that those who used informal
non-relative care were more likely to be EWEP or job club participants.
Childcare arra4ements did not generally vary according to the
demographic characteristics of the registrants, with one statistically
significant exception: Individuals with older children who could care for
themselves were more likely to have been employed during the two years
prior to their Initial program registration than registrants with younger
children. Additionally, registrants who used relatives to provide
childcare were less likely to have prior employment than individuals who
-123- 177
TABLE 5.6
SWIM
PERCENfAGE DISTRIBUTION OF CHILDCARE ARRANGEMENTSOF AFDC PARTICIPANT'S YOUNGEST CHILD.
BY PROVIDER AND LOCATION. AND BY REVIEW MONTH
Provider and Location Je:y 1986 November
Relative Provided Core 34.4 21.7In Child';, Hone 11.5 6.7In Provider's Home 14.8 10.0Location of Core Not Found in Cosefile 8.2 5.0
Non-Relative Provided Core 18.0 10.0In Child's Home .1.3 0.0In Provider's Home 8.2 8.3Location of Core Not Found in Cosefile 6.6 1.7
Group Core 1.6 6.7In Provider's Home 1.6 3.3in a Center 0.0 3.3
SWIM Activity Usually Occurred While Childwas in His/Her Regular School 0.0 23.3
Other Core Provided by a Relative 0.0 O
Other Core Provided by a Non-Relative 0.0 6.
Additional Core Provider Not Found in
Cosefile 0.0 16.7
SWIM Activity Always Occurred While Childwas in His/Her Regular School 3.3 10.0
Child Cored for Self (14 years or older) 29.5 16.7
Provider Information far July orNovember 1986 Not Found in Cosefile 13.1 11.7
Total 100.0 100.0
Sample Size° 61 60
SOURCE: See Table 5.5.
NOTES: Distributions may not add to 100.0 percent and subcategoriesnot odd to category totals due to rounding.
Tests of statistical significance were not examined.
1986
mcy
aFour participants (representing 7 children) who were active during
both July and November were randomly chosen for both the July and Novembersamples. Therefore. 4 participants and 7 children ale represented twice in thesample size.
-124-
TABLE 5.7
SWIM
PERCENTAGE DISTRIBUTION OF CHILDCARE ARRANGEMENTS OF AFDC PARTICIPANTS,BY PROVIDER OF CHILDCARE AND CHILD'S AGE
Provider
Child
Less Than6 Yeors
Child
6-9
Yeors
Child
1u-13
Yeors
Child
14 Yeors
or Older Total
Reiotive Provided Core 30.0 45.6** 55.4*** 0.0*** 33.9
Nan-Relative ProvidedCore 20.0 21.1 20.0 0.0*** 14.1
Group Core 30.0*** 8.8 1.5 0.0* 4.7
SWIM Activity UsuallyOccurred While ChildWos In His/Her
Regular School 20.0 8.8 16.9** 0.0*** 9.4
SWIM Activity AlwaysOccurred While ChildWas In. His/H,ir
Regular School 0.0 15.8*** 4.6 0.0 6.3
Child Cored for Self(14 yeci or older) 0.0* 0.0*** 1.54* 100.0** 31.8
Total 100.0 100.0 100.0 100.0 100.0
Sample Size° 10 57 65 60 192
SOURCE: MDRC coiculotions from cosefile reviews of randomly chosen AFDCregistrants who were active during July or November, 1986.
NOTES: Distributions may not odd to 100.0 percent due to rounding.
For each cell in the table, o statistical test was performed todetermine wiether the proportion of children in o porticulor age category using onorrongement was different from the proportion of children in all other oge groups whoused that orrongement. For example, the 30 percent of children less thon 6 years whohad relatives providing care is not sign_ficontly different from the combined percentin ether oge groups who used that orrongement. Differences qre statisticallysignificant using o chi - square test of the following levels: * = 10 percent; = 5percent; = 1 percent.
oTwenty-one children were excluded from this table due to missing
provider infornation.
-125-17D
used non-relative care.
Finally, the data yield limited inform tion concerning the extent to
which registrants utilized program monies to pay for childcare. Clear
evidence that program monies had been used for childcare during the
selected months was found in the case files of only 12 percent of the
participants. As already noted in Chapter 4, the 22 percent of the
participants whose children were all at least 14 years old were not
eligible for program childcare monies. The case files of 37 percent of the
participants showed no evidence of program-paid childcare. An unknowL
number of these registrants may have been only in activities for which no
childcare monies were available, fnr example, self-initiated education/
training or employment. For the remaining quarter of the sample, the data
did not allow any assessment to be made concerning whether program
childcare monies had been used during the two review months.
-126-1 SO
CHAPTER 6
OTHER PARTICIPATION ISSUES
One of the key features of SWIM was the requirement that registrants be
active in the program as long as they remained WIN-mandatory. Most other
welfare initiatives have had program requirements that were either limited
in duration or, if the program model specified ongoing participation, were
de facto short term. This chapter assesses the extent to which a
continuous participation requirement was implemented fn S 11 and provides
an overview of noncompliance activities and outcomes. In addition, the
chapter examines the extent to which registrants remained in the program
but escaped any participation withor,: being sanctioned.
I. Assessment of the Continuous Participation Requirement
The extent to which individuals participate in a progr-t on
continuous basis is a complicated concept to measure. In reading this
section, three measurement issues should be kept in mind. First, all
activities viewed by program staff as fulfilling the program's parti-
cipation requirements are incluck 1 in the definition of participation.
Thus, in addition to participation in program-arranged activities, employ-
ment while registered with SWIM and participation in seLf-initiated educa-
tion or training fall under the definition of participation, Second,
individuals were considered active in a month if they participated at least
one day during the month For example, no distinction is made between
those who participated one day and those who participated 20 days during a
-127- LS 1
month. Third, individuals were considered eligible for SW124 during any
month in which they were registered for ae least one day. In the two years
that SWIM operated, during a typical month over 75 percent of those
registered at least one day during the month were registered (and thus
SWIM-eligible) throughout the month (i.e., every day of the month). The
remaining 25 percent either registered or 4eregistared at sane point during
the month. This issue is discussed further in Chapter 7.
To examine continuous participation, the length of time that indi-
viduals were eligible for the program must be taken Ito account. The ten
hypothetical sample members shown in Figure 5.1 of Chapter 5 can be used to
illustrate this point.
Persons A and B were eligible for the program in all 12 months of a
one-year follow -up period. Person C, in contrast, was eligible for the
program in only three months -- October, November and December 1985 --
within a 12-N3nth follow-up period. For persons A and B, a continuous
participation requirement would involve 12 months of activity. For Person
C it would involve only three months of activity.
How long individuals remained registered with the program, and thus
eligible for program services, throughout a 12-month follow-up period is
shown in Table 6.1. Thirty-seven percent of the registrants were eligible
for SWIM throughout the 12-month period; 23 percent were SWIM- eligible for
one to three months out of the 12; 19 percent for. four to six mc:Ithr4 14
percent for seven to nine months; and 8 percent for ten or eleven months.
On average, registrants were eligible for the program for 7.8 months of the
12-month follow-up period.
The extent to which individuals
-128-
participated during every month they
182
TABLE 6.1
SWIM
PERCENTAGE DISTRIBUTION OF SWIM- ELIGIBLES BY PARTICIPATION STATUSAND NUMBER OF MONTHS REGISTERED FOR SWIM DURING THE
TWELVE MONTHS FOLLOWING .NITIAL REC, TRATION
Number of
Months Registered ParticipantsNon-
Participants Total
1 0.8 13.9 4.02 4.5 23.9 9.23 6.3 18.4 9.34 5.8 8.6 6.55 7.5 6.3 7.26 5.4 4.4 5.17 5.0 2.9 4.58 5.4 2.7 4.89 5.5 1.9 4.6
10 4.7 1.4 3.911 4.6 1.0 3.712 44.5 14.5 37.2
Total 100.0 100.0 100.0
Sample Size 1749 563 2312
SOURCE: MDRC calculations from the County of Son Diego Department ofSocial Services SWIM Automated Tracking System and EWEP attendance lags.
NOTES: The sample for this table includes SWIM -Eligibles whoregistered between Ju',y 1985 oll June 1986.
The sample is weighted to adjust for the higher proportionof AFDC's, relative to the proportion of AFDC-U's, in our sample.
Distributions may no odd to 100.0 percent due to rounding.
Participation is defined as attending EWEP far at least onehour or any other activity for at least one day.
Regis -eyed during a mantra is defined as registered for atleast oae day during the month.
Tests of statistical significance were not examined.
183-129-
were SWIM- eligible is shown in Table 6.2. Several points are evident.
First, as noted in Chapter 5, about one-quarter of the registrants
never participated within the 12-month follow-up period. Obviously, for
these individuals an ongoing participation requirement was not achieved.
It Shou,.) be noted that the majority of these nonparticipants were eligible
for the program for only a few months (see Tabla 6.1).1
Second, only a small proportion of all registrants -- 16 percent
were active during all the months in which they were eligible for pLogram
services. Inevitable periods of inactivity duo to assignment lags between
activities, illness, or other temporary interruptions in participation may
make this definition of continuous activity overly stringer' If con-
tinuous activity is defined as participating for at least 70 percent of the
months in which registrants were program eligible, then slightly over one-
third (36 percent) of all registrants were continuously active. Among
participants (i.e., registrants who ever participated during rho 12-month
period), almost 50 percent were continuously active using this definition.)
Appendix Table D.1 illustrates that these proportions did not seem to
vary according to the number of months individuals were resistered with the
program. Continuous participation was as likely for individuals with long
periods of eligibility as for those with short periods. In addition
(although not shown in the table), these proportions did not differ between
AFDC and AFDC-U registrants.
Those who were active in at least 70 percent cf their eligible months
participated, on verage, during 8.0 months (including employment as
participation) and during 4.9 months (not including employment as
participation), This average was virtually identical for AFDC and AFDC-U
-130-
Fi A
TABLE 6.2
SWIM
PERCENTAGE DISTRIBUTION OF SWIM-ELIGIBLES, BY PERCENT OFMONTHS ACTIVE OUT OF MONTHS REGISTERED, kiD BY ASSISTANCE CATEGCRY
Percent of Months ActiveOut of Months Registered AFDC AFDC-U Totol
100 15.8 15.2 15.6
90-99 5.5 4.7 5.2
80-89 7.3 6.5 7.0
70-79 7.6 8.8 8.0
60-69 8.9 6.4 8.0
50-59 11.8 14.2 12.7
40-49 3.8 2.7 3.4
30-39 5.9 5.7 5.8
20-29 4.9 2.4 4.0
10-19 3.9 5.3 4.4
1-9 1.4 1.8 1.5
0 23.3 26.3 24.4
Totol 100.0 100.0 100.0
Somple Size 1491 821 2312
SOURCE: MDRC colculotions from the County of Son Diego Depc tment ofSociol Services SWIM Automoted Trocking System ond EWEP ottendunce logs.
NOTES: The somple for this tobl e includes i ndivi duol s who registered
between July 1985 ond June 1986.
The sompie is weighted to ()Oust fdr the higher proportion ofAFDC's, relotive to the proportion of AFDC-U's, in our somple.
Distributions moy not odd to 100.0 percent due to rounding.
Porticipotlon is defined os ottending EWEP for ot leost onehour or ony other octivity for ot leost one 'Joy.
Registered during o month is defl.ed os registered ot leostone doy during he month.
Tests of stotisticol re not exomined.
8131-
registrants.
What types of activities typically occupiti the participants who were
continuously active according to this definition? Analysis of the program
tracking data indicated that among this group, employment while SWIM-regis-
tered was the most common activity, followed by education and training
activities, The mix of activities did not seem to depend on the number of
months an individual was registered with the program: These two types of
activities were most common among all individuals who were continuously
active, regardless of whether these individuals were eligible for SWIM for
a short or long time.
The fact that the program did not impose an ongoing participAtion
requirement for almost two-thirds of all registrants raises an important
question: Why were registrants inactive during months in which they were
program-eligible? This question is addressed briefly in the next section
of this chapter and in more detail in Chapter 7.
II. Noncom liance Activit and Outcomes
This section provides an overview of the extent to which registrants
failed to attend assigned activities, the activities in which this was most
likely to happen, and the eventual outcomes of these situations. Note that
the research definition of noncompliance differs from the Ae;Jlition used
by program staff. For research purposes, noncompliance is defined as the
failure of registrants to attend assigned activities or comply with other
program requirements. As is shown, many of those identified as noncom-
pliant in this study had valid reasons, from the program staff's point of
view, for not participating in or dropping out of a component.3
-132- ...
186
To describe the extent to which individuals were not in compliance
with program requirements, MDRC staff reviewed the case files of a subgroup
of registrants, both participants, and nonparticipants. The sample
consistek. of 144 AFDC and 98 AFDC-U SWIM - eligibles, randomly selected from
the group 'ho registered with SWIM between January 1 and March 31, 1986.
These registrEats' program activities were tracked for 15 to 18 months
after registration.
A summary of the findings based on the case file study is shown in
Figure 6 1 and Table 6.3. More than nine out of ten sample members (93
percent) were assigned to an active component -- either a pro3ram-arranged
activity, approved self-initiated education or training, or employment
while registered. A very high proportion -- four-fifths of dll registrants
(or 87 percent of those assigned to an activity) -- at sane point failed to
attend an assigned activity or failed to comply with sane other aspect of
program requirements. However, most of the noncompliant individuals had a
valid reason for this failure. Only 21 percent of all noncompliant
registrants were ever determined to have no legitimate reason for their
actions. Moreover, more than four-fifths of those with no legitimate
reason either eventually completed the activity in which they were noir-
campliant or were sanctioner..
The most frequent 4-ype of noncompliance was failure to initially show
up at an assigned activity (Table 6.4). Almost half of the noncompliant
registrants failed to attend their initially assigned job search workshop
session. One-fifth failed to attend their first job club session. Nine
percent failed to attend an EWEr orientation or an assigned EWEP wrksite.
Ten percent missed at least one appointment with program staff.
-133- 187
Figure 6.1
SWIM
FLOW OF SWIM-ELIGIBLES THROUGH THE NONCOMPLIANCE AND ADJUDICATION PROCESS
Not Assignedo an Active Component
18
Casefile StudySample
242
r
Assigned to anActive Component
224
Noncompliant at SomePoint With Program
Requirements194
Compliant With A7-1Program Requirements
30
Legitimate Reasons Establishedfor All Instances ofNoncompliance
153
Status After Noncompliance:b
Employedc
41Completed Activity 54Unknown 70
I
No Legitimate Reason Establishedfor At Least One Instance ofNoncompliance
41
1
Legitimate Reason forNo Legitimate Rervzon First Instance offor First Instanc' Noncompliance but Noof Noncompliance Legitimate Reason for
Subsequent Instance32d of Noncompliance
Completed Activityand NeverSanctioned 3
Completed Activitybut Sanctioned forSubsequent Instanceof Noncompliance 11
Did Not CompleteActivity andSanctioned 15
Did Not CompleteActivity and NeverSanctioned 3
1
Completed Activityand NeverSanctioned 3
Completed Activitybut Sanctioner, forSubsequent Insyanceof Noncompliance 0
Did hot CompleteActivity andSanctioned 2
nid Not Complete:tivity and Never
L rctioned 3
SOURCE: ANC calculations from casefile reviews of randomly chosenSWIM-Eligibles who registered between January 1 and March 31, 1986.
NOTES:oOf these 153 people, 98 had one instance of noncompliance; 55 had
two instances of noncompliance.bThe statuses after noncompliance are not mutually exclusive.
conly applies to those who became employed after the firstinstance of noncompliance.
d0f these 32 people, 6 had one instance of noncompliance and 26had two.
einformation regarding the completion status for the second
instance of noncompliance is missing for one of the nike people in thiscategory. it is known, however, that lnis individual was never sanctioned.
134 I 8
TABLE 6.3
SWIM
KEY NONCOMPLIANCE INDICATORS FOR SWIMELIGIBLES,BY ASSISTANCE CATEGORY
Indicator AFDC AFDCU Total
Assigned to an Active Component 93.8% 90.8% 92.6%
Noncompliant 79.9 40.6 80.2Reason for Noncompliance
No Legitimate Reason for at LeastOne Instance of Noncompliance 16.0 18.4 16.9
Legitimate Reason for All
Instances of Noncompliance 63.9 62.2 63.2
Sanct oneda
10.4 14.3 12.0
Total Number of SWIMEligiblesIn Casefile kcview 144 98 242
SOURCE: MC calculations from casefile reviews of randomly chosenSWIMEligibles who registered between January 1 and March 31, 1986.
NOTES: Noncompliance is defined as failing to attend an assignedactivity for the required number of days or failing to meet other programrequirements. Other requirements include verifying education or trainingattendance and behaving in a cooperative manner while attending a programcomponent.
Noncompliance indicators are calculated as a percentage ofall sample members in the indicated assistance category.
Tests of statistical significance were not examined.
°During the casefile reviews, a registrant was considered tobe sanctioned if there wa an indication in the registrant's program casefile thaf the grant had been reduced due to noncompliance with programrcquirements.
18 -3
135
TABLE 6.4
SWIM
PERCENT OF NONCOMPLIANT SWIA-ELIGIBLES.BY TYPE OF NONCOMPLIANCE AND ASSISTANCE CATEGORY
Activity and Type of NoncomplianCea AFDC .F0C-U TOTAL
Job Search WorkshopInitially Failed to Show Up at the ActivityInitially Showed Up of the Activity But
t'h'en Attendance Dropped OffType of Noncompliance Unknown
Job Club
Initially Failed to Show Up at the ActivityInitially Showed Up at the Activity ButThen Attendance Dropped Off
Showed Up at the Activity But Failed toMeet Other Types of Activity Requirements
EWEP
Failed to Att .ppraisal InterviewInitially Fa. . to Show Up at the ActivityInitially Showed up at the Activity ButThen Attendance Dropped Off
Disrupt,ve BehaviorType of Noncompliance Unknown
Education and TrainingInitially Failed to Show Up at the ActivityInitially Shawed Up at the Activity ButThen Attendance Dropped Off
Showed Up at the Activity But Failed toMeet Other Types of Activity Requirements
Type of Noncompliance Unknown
Failed to Attend Reappraisal Interview
43.5% 50.6% 46.41;
4.4 7.6 5.7
1.7 0.0 1.0
22.6 16.5 20.1
5.2 7.6 6.2
0.9 0.0 0.5
Sample Size
0.0 1.3 0.5
8.8 8.9 8.8
3.5 5.1 4.2
0.0 1.3 0.50.9 0.0 0.5
0.9 1.3 .0
7.0 2.5 f',.2
5.2 1.3 .6
0.I.' 1.3 .5
12.2 6.3 9.8
11S 79 194
SOURCE: MDRC calculations from casefile reviews of randomly chosen SWIM-Eligibles who registered between January 1 and March 31, 1986.
NOTES: Distributions may add to more than 100.0 percent because registrontrcould be nancompliant in mare than one program component.
(continued)
190-136-
TABLE 6..1 (continued)
Noncomplionce is defined os foiling to ottend on ossigned octivity forthe required number of cloys or foiling to meet other progrom requirements. Otherequirements include verifying educotion or troining ottendonce and behoving in ocooperotive monner while ottending o progrom component.
Tests of stotisticol significonce were not exomined.
ally the first two instances of noncomplionce for eoch individual orerecorded .a t. .s tobte.
19/
137
I: should be noted that the canponents in which registrants were most
commonly noncompliant were those which came early in the program model and
had the highest assignment rates. Thus, many individuals wen: noncompliant
In job search workshops, typically the first component to which a regis-
trant was assigned. A lacer proportion of indiv!.'uals were noncompliant in
program-arranged education or training, which typically occurred much later
in the model, than in self-initiated education and training, which could
occur earlier.4 When the extent of noncompliance in each program component
is calculated as the percentage of those assigned to a component who were
noncompliant in that component, the results indicate that the extent of
noncanpliance was relatively similar across components.
As noted in Chapter 4, registrants not in compliance with sane aspect
of the program were rescheduled for an activity (given a 'second chance')
and/or contacted by program staff before any formal actions were taken.
According to the case file reviews, 28 percent of the noncompliant
registrants were automatically rescheduled f.)r an activity at least once
without any in-person or telephone contalt between the registrant and
program staff. And is" percent of the noncompliant registrants were
rescheduled at leapt once, atter program staff initiated sane type ch.
contact with that through telephone calls. office meetings or hone visits.
Forty -five percent of the nonccmpliant registrants initiated contact with
program staff at least once to inform then of the reason for their absence.
Appenilix Table D.2 shows how often the different types of contacts were
made.
That many noncompliant individuals had legitimate reasons for their
behavior is confirmed by the case file reviews. Almost 84 1..ircent of the
-138- 1'92
noncompliant registrants were determined by program staff to have good
reasons for being unable or unwilling to participate in connection with
their first incident of noncompliance; 79 percent of the noncompliant
registrants were determined to have a !good cause,' for all identified
incidents of noncompliance.
For those whose first incident of noncompliance was legitimate, the
most frequently identified reasons were employment (25 percent of those
with good cause reasons), illness (13 percent), and denied or no longer on
welfare (11 percent). Other reasons included other appointments (6
percent), moved out of the SWIM area (4 percent), and waiting for a job to
begin (3 percent).
Among the noncompliant registrants who were determined to have no
legitimate reason for their actions, slightly more than 68 percent were
sanctioned at some point during the program.
Registrants had the right to contest a sanction decision at several
junctures in the formal adjudication proceedings, but very few (3 percent
of all noncompliant registrants) decided to do so.5 After a notice was
mailed to the registrant indicating an intent to deregister for noncompli-
ance, registrants had ten days to appeal the decision. Then, after
receiving a notice of an impending sanction, registrants had another ten
days in which to apnea1.6 If the registrant responded to this latter
notice, Income Maintenance (IM) staff would refer the matter to SWIM
program staff. If program staff decided to rescind the sanction request,
IM staff would comply; if not, IM staff would impose the sanction. (Note
that IM staff could not, based on their own judgment of the situation,
decide to ignore a request for a sanction.)
-130,3
The first sanction would affect a registrant's grant for three months;
a subsequent sanction would affect the grant for six months. In both
instances the effect on a registrant's check was determined by the
individual's assistance category and whether noncompliance had occurred in
EWEP. AFDC-U reaistrants who were noncompliant in an activity other than
EWEP lost all welfare benefits when sanctioned. AFDC-U registrants who
were noncompliant in EWEP had only their own needs removed frcm the grant
calculation. AFDC registrants noncompliant in any activity had their
grants reduced in the same way as AFDC-U's noncompliant in EWEP. For both
AFDC's and AFDC-U's, the registrant's Eligibility Technician would
generally notify the registrant in writing that he/she was again eligible
for welfare, once the sanction period had elapsed.
The adjudication process could be lengthy. Depending on whether an
individual responded to conciliation or contested the decision, the process
from the point when noncompliance was identified to the imposition of a
sanction -- could take from one to several months. For sanctioned
registrants in the case file review sample, for example, the number of days
from the first instance of noncompliance to the imposition of a sanction
ranged from 26 to 248 (averaging 124) days.
III. Coverage
The first section of this chapter investigatcZ how well SWIM
imp/emented an ongoing participation requirement. The results indicated
that approximately one-third of the sample was continuously active, defined
as active in at least 70 percent of the months in which they were
registered.
194-140-
NI1,
TJ..s section addresses a different question: To what extent did
program registrants remain in the program and escape both participation (of
at least one day) and sanctioning? This concept is referred to as
coverage; a program is said to *cover! all registrantt who either partici-
pate, are sanctioned, find employment, or leave welfare.
The longitudinal participation measures described so far were calcu-
lated for all individuals who registered for the program during SWIM's
first year of operation and took into account all program experiences with-
in 12 months after registration. This type of measure does not answer the
question of coverage for several reasons. First, simple calculations of
the percentage of individuals who participated during the follow-up period
may incorrectly suggest that all individuals remained eligible for SWIM
activities throughout the follow-up period and that large numbers of
individuals somehow avoided participation. In reality, as shown at the
beginning of this chapter, sane individuals remained eligible for only a
few months, leaving welfare or the program for a variety of reasons.
Additionally, as noted in the previous chapter, sane individuals were
penalized for their failure to participate by the imposition of a sanction,
that is, a temporary reduction in their AFDC grant. The extent of a pro-
gram's ability to reach the targeted caseload, and to enforce at least a
minimum participation requirement, is more evident when program eligibility
patterns and all program statuses are taken into account.7
The coverage indicators presented in this section establish the
proportion of individuals who, at a specific point in time after program
entry, were still on welfare and registered with the program, did not have
jobs, had never participated, and had not been sanctioned.
-141- 1 90
IN=M11W1W
The results indicate that very few individuals escaped this minimal
participation requirement in SWIM. At 12 months after registration, only 3
percent of both AFDC's and AFDC-U's in the sample had remained eligible for
SWIM, were not employed, had never participated in program-arranged
activities or self-initiated education or training, and had never been
sanctioned." (See Appendix Tables D.3 and D.4.)
Although direct comparisons are difficult to make, this degree of
coverage appears to be as high as or higher than that achieved by welfare
employment programs studied by MDRC as part of the Demonstration of Work/
Welfare Initiatives. 10
19C
-142-
CHAPTER 7
MONTHLY PARTICIPATION RATES
Testing the feasibility of having at least three-quarters of program-
eligibles active in a welfare employment activity at all times was, as
noted in Chapter 1, one of the basic goals of the SWIM demonstration. This
chapter examines the extent to which this benchmark was reached.
So far, the report has focu..ld on the typical program participation
patterns of registrants by tracking the experiences of an early group of
registrants for the 12 months following their initial registration. This
chapter uses a different analytical perspective and a larger sample of
registrants to address the following questions:
What percentage of the WIN-mandatory caseload was activeduring each calendar month of the demonstration?
During each month, in what types of activities wereregistrants participating?
What effect did the size of the monthly WIN-mandatory caseloadhave on the participation rates?
What effect does varying the definition of participation aswell as the el,:iinition of program-eligibles have on the rates?
Among registrants not active during particular months, whatwere the primary reasons for inactivity?
The chapter first introduces the concept of a monthly participation
measure. It then describes the samples used in calculating the monthly
participation rates. It goes on to examine monthly participation rates for
all 24 months of the SWIM demonstration, and to explore the effects of
calculating monthly participation rates in different ways. Finally, it
1. 9 7-143-
discusses the factors that constrain monthly participation rates, including
an assessment of Innat the SWIM results suggest about what the upper bounds
of an appropriate monthly participation (saturation) goal of a mandatory
welfare employment program might be.
I. Monthly Participation Measures
Monthly participation measures have been suggested as a means by which
to measure whether welfare employment programs successfully reach their
targeted caseloads with employment - related activities. In the past, these
measures have generally consisted of snap-shots of the numbers of individ-
uals participating or the types of statuses occupied by individuals in a
program during a certain period of time (for example, during a calendar
month, a quarter, or some other short period of time). Calculated as
rates, these measures take into account the number of individuals eligible
for the program within that period of time.
Figure 7.1, which depicts the program experiences of the same ten
hypothetical SWIM registrants used in previous illustrations, can be used
to illustrate how these rates are calculated. The ten individuals shown in
the figure registered at different points in the SWIM program and had
varying participation patterns after registration. The time period that
would be used to calculate a participation rate for the month of January
1986 is shown by a shaded area on the left side of the chart. Among the
ten individuals shown, only three -- persons A, B and D -- were registered
with the program during this month. Two of these individuals -- persons B
and D -- were active during this month, resulting in a monthly participa-
tion rate of 67 percent for this snail sample. This is the type of
-144- 1 9 0-,
Person A
Person B
Person C
Person D
Person E
Person F
coPerson G
Person H
Person I
Person J
199
FIGURE 7.1
SWIM
LLUSTRATION OF TIME PERIODS COVERED BY MONTHLYPARTICIPATION RATES USING TEN HYPOTHETICAL SAMPLE MENEERS
1985 1986
NovL. 1987
Jul Sep1
Nov Jan
.
Mar
l
May Jul,
SepI
Jan Mar May1 , I ... .....i
I 1.11-1-1--I
H IH'
I I IIH
I
I F--IH
- 7'
1
FH IH'
I--'Hi--i
Registered and
Participating
Registered andNot Participating
Time Period Usedin Calculating Rates
ri
20.)
participation measure presented in this chapter, estimated for all 24
months of the demonstration.
The concept of examining participation levels on a monthly basis is
familiar to program managers -- in terms of the number of individuals
participating, although not expressed as a rate -- as a way of indicating
how a program is likely to look at any given point. This knowledge informs
managers about the use of various program activity 'slots,.' on a monthly or
quarterly basis. Fran the evaluator's point of view, monthly participa-
tion measures, calculated at set intervals, indicate the array of rendered
program services at any point in time and ha these services change over
time. Early in a program, sane components may not be fully implemented or,
in a sequential program model, registrants may take a long time to reach
later components. Monthly participation rates, disaggregated by type of
program component, can highlight these factors.
Five limitations to this type of measure should be noted, however.
First, unlike the longitudinal measures presented in Chapters 5 and 6,
monthly participation rates cannot characterize the flow or path of
registrants through a program by depicting typical sequences of activities;
nor can they indicate how long individuals participated in the program and
whether participation was on a continuous basis.
Second, these rates examine participation in a program as of a certain
point in time, ignoring the prior and subsequent program experiences of
registrants; the rates do not indicate whether the same individuals are
participating month after month or, conversely, whether the same
individuals are inactive month after month.
Figure 7.1 can also be used to illustrate this point. If a monthly
-146-201
participation rate for the month of October 1986 (highlighted by the middle
shaded area in the figure) were calculated for the ten hypothetical sample
members, persons A, B, F, G and H would be considered eligible for the
program and all except person H would be considered active. The monthly
participation rate for this month would be 80 percent. But this rate would
not indicate that as of October 1986, Person A was ending a year-long
period of intermittent program participation; that Person B was in the
midst of a 21-month period of continuous participation; that Person F was
beginning a medium-length participation spell; that Person G was ending a
short interval of participation; and that Person H never participated in
the program, even though he/she was registered with the program for several
months.
(A later section of this chapter evaluates the extent to which those
who did not participate in any given month were the same individuals month
after month. For this analysis, a different type of participation measure
is used to address the issue of frictional inactivity.)
The third limitation of monthly participation rates is that they
generally measure actual participation, not compliance with program
requirements.1 As described in previous chapters, many individuals were
not active at certain points during their program tenure with good reason.
Registrants often waited several weeks for a scheduled component to begin;
some individuals were inactive for short periods of time due to illness;
and a small number (undocumented workers, for example) were never assigned
to a program component. (The last section of this chapter examines the
situations of individuals who did not attend program activities during
selected months and estimates the extent to which individuals were
-147- e202
complying with program requirements during each month, even if they were
not attending program activities.)
Fourth, monthly participation rates do not include any individuals who
were not eligible for the program in a given month because they had been
sanctioned for noncompliance with program requirements and their sanction
was in effect.
Finally, it is important to note that the preceding illustration and
explanation of monthly participation rates belies the complexity of actu-
ally calculating this type of rate. Complicated computer programming is
necessary to take into account the various eligibility and participation
patterns of registrants. Computer programs, for example, must allow for
the fact that individuals may be eligible for a program for only part of a
month, and may participate in several different kinds of activities during
a month, not all of which necessarily may count as participation.
Additionally, monthly participation rates are very sensitive to the quality
of the eligibility and participation information used to calculate the
rates. For example, the dates recorded for the beginnins and end of a
period of participation or eligibility can have a big effect on how high
the calculated rates are, making accurate recording extremely important.
II. Samples Used to Calculate Monthly Participation Rates
The participation statistics presented in Chapters 5 and 6 were
calculated for those individuals who registered with SWIM during its first
year of operation. For this group -- the same individuals for whom impacts
are presented in Chapter 8 -- the statistics took into account program
activities that occurred during any month within 12 months of registration.
-148-
203
To calculate monthly participation rates over the cnurse of the demon-
stration, all individuals who were eligible for SWIM services during the
24-month research period must be represented in the sample.
To understand this difference, look again at Figure 7.1. For the
types of rates calculated in Chapter 5, Persons A, B, C, D, E and F would
be included, since they registered with SWIM during the program's first
year and thus have 12 months of follow-up available. For a December 1986
monthly participation rate, Persons B, F, H, I and J would be included,
since they were eligible for the program in that month. (See the shaded
area on the right side of Figure 7.1.) These five individuals did not
necessarily register with SWIM during the program's first year of
operation. In fact, these five individuals registered with SWIM at various
points between September 1985 and November 1986.
The number of registrants eligible for SWIM services in each month of
the demonstration -- that is, the number of individuals who were registered
with the program as of each month -- is shown in Table 7.1. Sane indi-
viduals, of course, will be represented in several or all months. The
group of individuals eligible for the program in any given month is
referred to as the monthly registrant caseload.2
Several points can be made from Table 7.1. First, the fact that the
first 12 months of SWIM represented a phase-in period becomes very clear.
During these months, individuals who were WIN-mandatory before the start of
SWIM were phased into the program. In addition, any individuals who
applied for welfare or recipients who were newly determined to be WIN-
mandatory during the first 12 months of SWIM became eligible for the
program once they registered. Most frequently, those who were newly
-149- 2 04
TABLE 7.1
SW IM
NUMBER OF REGISTRANTS ELIGIBLE FOR SW IM SERVICES,BY MONTH AND ASSISTANCE CATEGORY
Month AFDC AFDCU Toto I
Jul y 1985 338 184 522
August 617 329 946
September 802 420 1222
October 1034 548 1583
November 1205 609 1814
December 1287 684 1971
Jonuory 1986 1408 761 2168
Februory 1551 800 2351
Morch 1692 854 2546
April 1761 884 2645
May 1816 936 2752
June 1959 982 2942
July 2114 1059 3173
August 2203 1104 3307
September 2296 1109 3406
October 2355 1133 3488
November 2326 1142 3468
December 2422 1191 3613
January 1987 2475 1229 370.1
Februory 2512 1233 3745
Morch 2575 1239 3814
April 2616 1257 3873
Moy 2527 1219 3746
June 2543 1225 3768
Sompl e Size 5332 2949 8281
150_213
(continued)
TABLE 7.1 (continued)
SOURCE: MDRC calculations from the County of San Diego Department of SocialServices SWIM Automated Tracking System.
NOTES: The sample for this table includes individuals who registeredbetween July 1985 and June 1987.
The sample is weighted to reflect the actual number of
The number of AFDC and AFDCU registrants may not sum to totals due
Eligible during a month is defined as registered for at least oneday during the month.
SWIMEligibles.
to rounding.
151
206
determined to be WIN-mandatory were former WIN-exempt recipients whose
youngest child had recently turned six years of age.
After the end of the 12-month phase-in period, as Table 7.1 indicates,
the monthly registrant caseload continued to grow for several months,
reflecting the fact that the numbers of new registrants (applicants and
redetermined recipients) were outweighing the numbers of individuals
deregistering from the program. This could reflect seasonal fluctuations
in welfare applications and case closures or an overall increase in welfare
applications.
Beginning in December 1985, however, the registrant caseloads
stabilized. For the next seven months, between 3,600 and 3,900 individuals
were eligible for SWIM services in any given month. Also evident from
Table 7.1 is that AFDC-U's comprised about a third of the registrant
caseload in any given month.
The phase-in of the current WIN - mandatory caseload during the first 12
months of SWIM affected the composition of the registrant caseloads during
the first year of the program as well as the size of the caseloads. In
terms of both prior employment and prior exposure to the program, the
registrant caseloads during the second year of the program were more
typical of the ongoing monthly registrant caseloads of the two SWIM
offices.
Compared to the registrants eligible for SWIM during later months of
the program, the registrant caseloads in the early months of SWIM consisted
of individuals who had less prior employment. For example, among those
eligible for SWIM during December 1985, 42 percent had not worked at all
during the two and one-half years prior to this month, 49 percent had
-152-
207
worked in one to eight quarters during this time period; and 9 percent had
worked almost steadily -- in nine or ten quarters -- during these two and
one-half years. This distribution reflects the predominance, in the early
months, of individuals who were phased into the program because they were
WIN-mandatory prior to SWIM. This group probably included many long-term
welfare recipients.
As applicants and redetermined individuals registered with the
program, and as sane of the previously WIN-mandatory individuals
deregistered, the registrant caseload in any given month became less
!disadvantaged.' For example, the employment histories of those eligible
for SWIM during February 1987 reflect a greater extent of prior employment.
Among these individuals, only 35 percent had not worked at all during the
two and one-half years prior to February 1987; 56 percent had worked ln one
to eight quarters during this period; and 9 percent had worked almost
steadily -- in nine or ten quarters -- during these two and one-half years.
The registrant caseloads in different months of program operation also
reflect varying lengths of stay in SWIM. Table 7.2 indicates the 'program
history' for those eligible for SWIM in each month of the demonstration.
For example, among those eligible for the program in December 1985, no one
could have been in the program for more than five months, since SKIM did
not start until July 1985. In fact, 30 percent had been registered for
four or five months; 54 percent had been registered one to three months;
and 16 percent were new to the program that month. Given the sequential
nature of the program model, few of these individuals would be expected to
be in the program-arranged education or training portion of the model;
most would be expected to be in job sear "h or EWEP. By February 1987, in
-153- 2 0
TABLE 1.2
PERCENTAGE DISTRIBUTION OF INDIVIDUALS REGISTERED WITH SWIM AS OF EACH MONTH,
BY NUMBER OF MONTHS REGISTERED WITH THE PROGRAM PRIOR TO EACH MONTH
Nir of Prior
Reg1.3tereC Months
1985 1986 1987
JUL AUG SEPT OCT NOV DEC JAN FEB MAR APR MAY JUNE JULY AUG SEPT OCT NOV DEC JAN FEB MAR APR MAY JUNE
Registered During
the month
1-3
4-6
7-9
10-12
13.45
16-18
19-21
22-23
Number of indivi-
duals Registered 522 946 1222 1583 1812 1971 2168 2351 2546 2645 2752 2942 3173 3307 3406 3488 3468 3613 3704 3745 3814 3873 3746 3768During the month
109.0 46.7 32.0 30.5 19.8 16.4 16.6 15.8 14.8 11.5 11.9 12.4 13.4 11.6 10.7 9.5 7.4 8.1 8.4 6.8 8.1 8.1 3.9 1.1
53.3 68.0 69.5 61.6 54.1 46.3 40.0 37.0 37.1 31.9 29.1 27.9 29.5 29.2 21.6 25.4 23.4 21.2 20.6 19.1 19.1 19.3 16.5
18.6 29.5 37.1 33.2 30.4 27.8 26.5 24.6 23.8 21.0 19.4 19.6 21.2 20.3 20.4e18.9 17.4 15.1 15.1 14.1
11.0 17.9 23.6 22.3 21.2 19.1 17.7 17.7 17.0 16.3 15.1 15.4 16.5 16.2 16.4 16.4 14.5
7.3 12.7 15.9 15.4 14.8 15.3 13.8 14.4 13.8 13.1 12.3 12.9 14.0 13.2
4.8 8.2 11.0 12.1 11.9 11.6 10.9 11.4 11.3 11.0 11.1
3.7 6.8 9.2 9.9 9.7 9.6 9.2 9.9
3.2 5.8 7.1 8.3 8.6
2.7 5.0
SOURCE: LWRC calculations from the County of Son Diego Department of Social Services SWIM Automated Tracking System.
NOTES: The sample for this table Includes individuals who registered between July 1985 and June 1987.
The sample Is weighted to reflect the actull number of SWIM- Eligibles.
Distributions may not add to 100.0 percent due to rounding.
Registered during a month is defined as registered for 9011A at least one doy during the month.
contrast, large groups of Individuals had been in the program for long
periods of time. Among those in the registrant caseload for this month, 24
percent had been registered at least 13 months; 30 percent had been
program-eligible for seven to twelve months; 40 percent for one to six
months; and 7 percent were new to the program. The majority of these
individuals would be expected to be in the education/training portion of
the model. A smaller portion would be expected to be in job search or
EWEP. It should be kept in mind that in terms of prior exposure to the
program the registrant caseload in an ongoing program operating in these
two SWIM offices would probably more closely resemble the 1987 than the
1985 registrant caseloads.
III. Monthly Participation Rates in SWIM, Varying the Types of ActivitiesCounted as Participation
Monthly participation rates are broadly defined as the number of
individuals active during a month divided by the number of individuals
eligible for the program during that same month. The magnitude of the
resulting rates can be affected by several factors. This section discusses
how counting different types of activities as participation can affect the
monthly participation rates. As becomes clear, most striking of all is the
role that employment and self-initiated education and training activities
play in the monthly participation rates.
Figure 7.2 shows, for each month of the demonstration, the proportion
of individuals eligible for SKIM (AFDC registrants combined with AFDC-U
registrants) who participated in program-arranged activities, self-
initiated activities and/or employment while registered. 3 Individuals in
-155 -
211
FIGURE 7 . 2
SWIM
PERCENT OF INDIVIDUALS ELIGE3LE FOR SWIM DURING EACH MONTHWHO PARTICIPATED IN PROGRAMARRANGED ACTIVITIES. SELFINITIATED
ACTIVITIES OR EMPLOYMENT. BY TYPE OF ACTIVITY
JUL85
2 J 2,
OCT85
JAN86
APR86
rlEmployment
71Self-hitiatedActivitiesIIProgram-Arranged
Acbvities
JUL86
Month
OCT JAN86 87
APR87
213
any csi these activities were considered, by the program, to be fulfilling
program requirements. (See Appendix Table E.1 for a breakdown of these
rates by component and assistance category for two typical months of the
demonstration.) Note that individuals were considered eligible for SWIM
during a month if they were registered with the program for at least one
day in the month. Participation was defined as attending an activity or
working for at least one day in the month.4
Several points can be made from this figure. If only program-arranged
activities are considered participation, 26 to 39 percent of the
registrants eligible for the program in any given month during the first
year cf SWIM would he considered active. In this first year, however, the
current WIN-mandatory caseload was being phased into the program and
monthly registrant caseload sizes were low. During the second year of
SWIM, when the caseloads were larger and more closely resembled those of an
ongoing program, monthly participatior rates counting only program-arranged
activities were lower. In any gi-ren month, betty 18 and 28 percent of
those eligible for the program were participating in program-arranged
activities. 5
Adding self-initiated education and training to the participation
criteria results in monthly participation rates of between 26 and 42
percent during the first year of SWIM, and between 31 and 35 during the
second year. Note that activity includ:- here reflects participation in
registrant-initiated education or training that was both known to and
approved by program staff. Again, the rates achieved during SWIM's second
year more accurately reflect how SWIM would operate as an ongoing program.
The most noteworthy feature of the self-initiated activity is that,
-157-
21
over time, an increasing proportion of each month's registrant caseload was
active in self-initiated education and training. In October 1985, 3
percent of the registrant caseload was in self-initiated education or
training; by July 1986, this proportion had increased to 7 percent; by
January 1987, to 13 percent; and by May 1987, to 14 percent. As discussed
below, two possible factors could explain this pattern. First, over the
course of the demonstration, registrants may have become increasingly aware
of the fact that self-initiated activities could defer them from
program-arranged components. Second, registrants who became active in
self-initiated education or training programs may have remained
SWIM-eligible, while enrolled in these programs, for long periods of time.
Adding employment while registered to the activities included in
participation results in monthly participation rates of 35 to 59 percent
during the first year of program operations, and between 47 and 55 percent
during the second year of SWIM. These rates reflect the fact that after
the first several months of the program, approximately 18 percent of the
SWIM-eligible individuals in any given month were employed and did not
participate in any other activity.
This more comprehensive definition of participation -- which includes
program-arranged activities, self-initiated education or training known to
and approved by program staff, and employment while registered with the
program -- is the one most commonly used In this chapter. This is because
this definition most closely represents all activities that fulfilled SWIM
program requirements.
As mentioned above, the proportion of registrants active in program-
arranged activities decreased over time. Figure 7.3 shows, for each month
-158-
2.1 6
.0.
FIGURE 7 . 3
SWIM
PERCENT OF INDIVIDUALS ELIGBLE FOR SWIM DURING EACH MONTH WHOPARTICIPATED IN PROGRAMARRANGED ACTIVITIES, BY TYPE OF ACTIVITY
ElEducation or TrainingProgram-Arranged
Ei Job SearchActivities
IIIWort( Experience
JUL85
216
MI IIIIOCT85
JAN86
APR86
Month
JUL86
OCT86
JAN87
APR87
217
1
I
of the demonstration, the proportion of individuals eligible for SWIM who
participated in any of the three broad categories of program-arranged
activities; work experience, consisting almost totally of EWEP; job search
activities, including job search workshops, job clubA and ISESA; and
program-arranged education or training.
The two factors that contribrted to the decline in the program-
arranged monthly participation rates can be clearly seen. First, the
proportion of individuals in job search and, to a lesser extult, the
proportion in BM, decreased over time. Second, the proportion of
individuals in program-arranged education or training remained fairly
constant.
The first factor is to be expected. Individuals who registered for
SWIM would be expected to be involved in job search and EWEP during the
first four to eight months they were in the program. After this point, if
an individual were ctill eligible for the program, he/she would be referred
to an educatif ,Ing program, Thus, depending on had long indi-
viduals remaineL .ligible for SWIM, one would expect to see the proportion
of individuals in EWEP and job search decreasing over time.
The second factor is the surprising one. As noted, if individuals
were still eligible for SWIM, the expected step after four to eJ.ght months
would be referral to education or training. 6 And, as iloted in an earlier
section of this chapter, the registrant caseloads in the second year of
SWIM consisted of many individuals who had been registered with the program
for a while. These individuals should have been active in the education
And training portico of the model during the later months of SWIM, thereby
increasing the mont'7. share of program-eligibles in these types of
-160-2, 1 z)- .r i.)
components.
There are several possible reasons for the unexpected pattern. The
first explanation for the trends depicted in F.gure 7.3 focuses on the grow-
ing predominance of self-initiated education and training over the course
of the demonstration. Several things could account for this. For one,
interviews with program staff indicated that over the course of the
demonstration, registrants became increasingly aware of the fact that self-
initiated activities would defer then from program-arranged components.
The numbers presented in Chapter 5, however, indicate that only a snail pro-
portion of registrants -- 5.5 percent of AFDC's and 4.3 percent of AFDC-U's
-- enrolled themselves in education or training programs after initially
participating in another activity. The more likely explanation, therefore,
is that registrants who were active in self- initiated education or train-
ing programs may have remained SWIM- eligible, while enrolled in these
programs, for long periods of time.
The second possible explanation concerns the fact that as the current
WIN-mandatory caseload was phased into the program, and as more and more
individuals finished the job search and EWEP portion of the SWIM model, CRU
job development counselors may not have referred all eligibles to education
and training, as the program model assumed. Their caseloads were growing
rapidly; and, as noted earlier in the report, interviews with those staff
during the last few months of the demonstration indicated that several of
them, in reaction to their large workloads, decided not to work with all
post-assessment registrants in their caseloads, although they had reported
doing this successfully up to that point.
The increase in the number of individuals eligible for SWIM relative
to the number of individuals participating is shown in Figure 7.4. Note
that, in this figure, components counting towards participation include
program-arranged activities, self-initiated activities and employment while
registered. As indicated earlier in the chapter, the number of registrants
eligible for SWIM increased over time. This number was highest in April
1987, when 3,873 individuals were registered for at least one day (see
Table 7.1). The number of individuals participating in any given month did
not keep pace with the growing caseloads. Indeed, this number increased
only slightly during the second year of SWIM. At most, again during April
1987, 2,099 individuals were participating.
Overall, however, as noted above, participation rates, including all
types of activities, remained fairly constant -- hovering between 47 and 55
percent during the second year of SWIM. This indicates that the share of
registrants available for program-arranged activities might have declined
over time, reflecting the fact that increasing portions of the monthly
registrant caseload were in self-initiated activities or were employed.
IV. Monthly Participation Rates in SWIM for AFDC's Compared to AFDC-U's
The preceding section presented monthly participation rates in SWIM
for all those eligible for the program in each month. This section
presents monthly participation rates calculated separately for AFDC and
AFDC-U registrants.
Three types of monthly participation rates are presented for AFDC and
AFDC-U registrants in Figure 7.5, reflecting differences in the definition
of participation. For two out of the three definitions -- participation
defined to include only program-arranged activities, and participation
- 52-
ati
FIGURE 7 . 4
SWIM
NUMBER OF INDIVIDUALS ELIGE3LE FOR SWIMAND NUMBER OF INDIVIDUALS PARTICIPATING IN SWIM,
DURING EACH MONTH
i,0N(...)
1
i:ft-E
..Q
.4.
4000
3000
2000
1000
0 IIIIIIIIIIIJ___L___J.__I___L_IIIIIIII
Total Number of Individuals Eligible
_,- for SWIM During Each Month
..........------...---
Total Number of IncivickialsWho Participated At Least OneDay Duriv Each Month
JUL OCT JAN APR JUL OCT JAN APR
85 85 86 86 86 86 87 87
Month
2r2,1222
FIGURE 7.5
SWIM
PERCENT OF INDIVIDUALS ELIGIBLE FOR SWIM DURING EACH MONTHWHO PARTICIPATED IN PROGRAMARRANGED ACTIVITIES,
SELFINITIATED ACTIVITIES OR EMPLOYMENT,BY ASSISTANCE CATEGORY AND TYPE OF ACTIVITY
........
JUL
85
2 J
OCT85
AFDC
AFDCU
Program-Arranged Activities.Self-Initiated Activitiesand Employment
Program-Arranged Activities. ....... and Self-Initiated Activities
L____I L I L I I
JAN APR JUL OCT JAN APR
86 86 86 86 67 87
Month
Program-Arranged Activities
224
defined to include program-arranged activities as well as self-initiated
activities -- the monthly participation rates for AFDC's and AFDC-U's are
very similar. Within these broad definitions of participation, the program
components in which these two groups participated were also similar,
although AFDC-U's were slightly more likely than AFDC's to participate in
job search, and slightly less likely to participate in education or
training. (See Appendix Table E.1.)
When the definition of participation is changed to include employment,
however, the monthly participation rates for AFDC's are generally higher
than those for AFDC-U's. This difference is most probaty due to the '100
hours' rule in effect for AFDC-U's, under which AFDC-U's who work more than
100 hours in a month become ineligible for welfare. This rule does not
apply to AFDC's. Thus, including employment as an activity counting
towards monthly participation has more of an effect on AFDC monthly
participation rates than on AFDC-U monthly participation rates.7
V. Monthly Partici ation Rates in SWIM Var in the Definition ofProgram-Eligibles
So far in this chapter, individuals eligible for the program in any
month have been defined as those who were registered with SWIM for at least
one day during that month. This is not the only way of defining which
individuals were eligible for the program during each month of program
operations. Other ways include defining program eligibles as individuals
who were registered throughout the month or as individuals registered as of
the beginning or end of a month.
To illustrate this point, Figure 7.6 shows an enlarged version of the
-165-
2'd
2 2- 6'
Person A
Person B
Person C
Person D
Person E
Person F
Person G
Person H
Person I
Person J
FIGURE 7.6
SWIM
ILLUSTRATION OF TIME PERIOD COVERED BY A MONTHLY PARTICIPATIONRATE FOR NOVEMBER 1986 USING TEN HYPOTHETICAL SAMPLE MEMBERS
1986
Oct Nov
Registered and
Participating
I
Registered andNot Participating
Time Period Usedin Calculating the Rate
2 2 7
November 1986 column of Figure 7.1. Sample members' activities during this
month are shown in the shaded area. Defining program eligibles as those
who were registered throughout the month would !flag' Persons B, F and H.
Of these three individuals, two were active during the month, resulting in
a monthly participation rate of 66 percent. Defining program eligibles as
those who were registered as of the end of the month would flag Persons B,
F, H, I and J. Two of these five individuals were active, resulting in a
monthly participation rate of 40 percent. Defining program eligibles as
those who were registered at least one day during the month (the definition
used so far in this chapter) would flag persons A, B, E, F, H, I and J as
program eligibles. Three out of these seven registrants were active,
yielding a monthly participation rate of 43 percent. As evident by this
illustration, changing the definition of program-eligibles changes both the
numerator and denominator of monthly participation rates.
When the most comprehensive definition of participation is used,
monthly participation rates vary between 3 and 11 percentage points each
month, depending on how program-eligibles are defined (Figure 7.7).8 As
expected, the direction of the difference is consistent in every month:
Participation rates are highest for those registered throughout t%e month.
Rates including only those registered as of the end of each month are next.
Monthly participation rates which define eligibles as those who were
registered at least one day during the month are the lowest.
These differences are caused by the fact that changing the definition
of program-eligibles in this way increases the denominator of the rates
while decreasing the amount of time during which the program can work with
an individual.9 For example, some individuals were eligible for the
-167 -
228
223
100
50
FIGURE 7.7
SWIM
PERCENT OF INDIVIDUALS ELIGBLE FOR SWIM DURING EACH MONTHWHO PARTICIPATED IN ANY ACTIVITY, USING DIFFERENT EUGBIUTY CRITERIA
0 i i 1 1 1 1 1 1 1 1 1 1 1 1 1 I 1 1 1 1 1 1 1 1
AL OCT JAN APR JUL OCT JAN APR
85 85 86 86 86 86 87 87
Month
Eighty Defined as Registered
Throughout Month
Eigiality Defined as Registered
at End of Month
Eigibitty Defined as Registered
at Least One Day Durhg Month
230
111111111=1111111MMIIIIIIIIIIIMINIIIIW
program for only a short period of time during a month. Some of these
short-term eligibles probably did not receive services because they were
registered with the program for too short of a time; applicants denied
welfare would fall into this category. Another group of individuals
probably registered with the program during th,a month. Some of these new
registrants would be assigned to activities that were scheduled to begin
during the following month.
The 3 to 11 percentage point differences in these rates for the SWIM
program do not seem to warrant much concern in and of themselves. However,
if the continuation of a program depended on the achievement of a specific
rate, or funding decisions were affected by participation rates, this small
difference could became important.
It is likely that the effect of changing the definition of program-
eligibles would depend, in part, on the type of program operated. In a
program with a high registrant turnover rate (e.g., a program targeted only
toward welfare applicants), which conducts extensive assessments when
individuals .nitially enter the program, defining program-eligibles as
those eligible for the program throughout the entire month may yield the
highest rate. In this type of program, a large group of registrants may
stay in the program for only a short period of time and, due to the
extensive assessment activities at the beginning of the program model, may
never reach the participation stage of the model. Defining program -
eligibles as those registered at least one day during the month would put
this type of program at a disadvantage in meeting monthly participation
targets, relative to programs that did not have such pre-participaticin
components and such high turnover.
-16 9-
23i
In programs like SWIM -- where the targeted population included the
entire WIN-mandatory caseload and individuals were immediately assigned to
job search at registration -- defining program - eligibles as those
registered at leapt c'e day during each month seems appropriate. Only a
small proportion -- approximately 9 percent -- of each monthly registrant
caseload during the second year of SWIM consisted of individuals who
registered with the program during the month. Thus, rz;istrant turnover
was not high. Also, most of these new registrants would be expected to
participate within a short time of registration.
Fran a practical point of view, the type of program activity tracking
systems available in a locality would determine the ease with which an of
these rates could be calculated. Defining program-eligibles as those who
were registered throughout a month would generally be the most difficult of
all to calculate, since this would require knowledge of registrants'
eligibility status as of every day of the month.
VI. Participation Rates in SWIM, Varying the Period of ime over WhichRates are Calculated
So far in thi, chapter, participation rates have been presented as
monthly rates. The period of time over which rates are calculated can, of
course, vary. Other plausible time periods are a day (e.g., the last day
of each month) or a quarter.
It is useful to illustrate the different rates using the same hypo-
thetical sample members as depicted in Figure 7.5. A participation rate
calculated as of the last day of October 1986 would result in a rate of 75
percent: Three of the four individuals eligible as of that day were active
-170-
232
as of that day. A participation rate calculated for the month of October
1986 would yield a rate of 80 percent: Four of the five individuals who
were eligible for the program at same time during the month were active
during the month. For the hypothetical sample members, participation rates
calculated for the months of October, November and December 1986 would be,
respectively, 80 percent, 43 percent, and 40 percent. A participation rate
calculated for the three months together (i.e., for the last quarter of
1986) would produce a rate of 50 percent: Four of the eight individuals
who were eligible at any point during the quarter were active during the
quarter.
Quarterly participation rates in SWIM are compared with monthly parti-
cipation rates in Figure 7.8. (Note that the quarterly rates are graphed
at the middle month of each quarter.) For both types of rates, participa-
tion included program-arranged activities, self-initiated education or
training, and employment. As can be seen, quarterly rates in SWIM were
higher than monthly participation rates. This is attributable to the fact
that registrants were more likely to participate during at least one of the
three months in each quarter than in one specific month. Although not
depicted in the figure, this pattern held true regardless of whether self-
initiated activities or employment were counted as participation.10 The
difference between the quarterly rate and the corresponding monthly rate
within each quarter ranged between 4 and 19 percentage points.
As with the differences in monthly participation rates according to
the definition of program-eligibles, the generally amdll difference between
these two types of rates does not seem troublesome in and of itself. Nor
does one type of rate necessarily yield more information about the program
-171- 233
100
FIGURE 7 . 8
SWIM
PERCENT OF INDIVIDUALS ELIGELE FOR SWIM DURING EACH QUARTEROR MONTH WHO PARTICIPATED IN ANY ACTIVITY
JUL
85OCT85
JAN86
Quarterly Participation Rates
Monthly Participation Rates
I 1 I 1 j___L 1 l I I 1 1 1 1___1
APR JUL OCT JAN APR86 86 86 87 87
Month
If 234 235
than the other. However, if policy decisions are to be based on the
achievement of a specific participation rate, even the small difference
between the two could become important.
VII. Reasons for Nonparticipation During Selected Months
As the preceding sections of the chapter indicate, depending on the
way the rate is calculated, approximately half of those eligible for SWIM
in any given month actually participated in some type of activity or were
employed during the month. This section examines those who were not active
during selected months and analyzes possible reasons for nonparticipation
during these months. The results .shed light on the factors that constrain
monthly participation rates and the issue of whether the maximum partici-
pation levels possible in the program were achieved.
This section addresses three separate questions: First, to what
extent are those categorized as inactive lie same individuals each month?
That is, did a certain group of individuals remain eligible f- the program
for a long period of time and never participate? 11 Second, do the demo-
graphic characteristics of inactive registrants liffer fray, those of
participants? Third, for what reasons did individuals not participate?
A. Turnover Among Inactives
To examine whether the same individuals were inactive month after
month, one must first determine the extent to which the same individuals
were eligible for SWIM month after month, and then determine the extent to
which these continuously registered individuals never participated.
The data indicate that 1,411 individuals were continuously registered
during the second year of SWIM (i.e., registered with the program for at
-173-
236
least one day during every month between July 1986 and June 1987). These
1,411 individuals represented about two-fifths -- between 36 percent and 45
percent -- of the number o2 SWIM - eligible registrants in any given month
during this period. The data also inlicate that only 266 of these indi-
viduals did not participate at all during the second year of SWIM. These
266 individuals represented between 13 and 19 percent of the EWIM-eligibles
in any given month who did not participate in that particular month.
Thus, registrants eligible for the program but inactive during any
given month were generally not the same individuals month after month:
Less than half of those eligible for the program in any given month were
eligible for the program throughout the second year of SWIM. In addition,
less than one-Lifth of these continuously eligible individuals did not
participate at sane point during the 12-month period.
B. Demographic Characteristics of Nonparticipants in Selected Months
In order to identify what types of registrants were inactive, the
demographic characteristics (measured at initial program registration) of
nonparticipants were ccmpared to those of participants among those eligible
for SWIM for at least one day in July 1986, or November 1986. Participation
included program-arranged activities, self-initiated education or training,
or employment.
Participants and nonparticipants differed along several key
dimensions. (See Table 7.3.) The most striking feature: of the comparison,
apparent . both the July and November samples, is the preponderance of
applicam t among the nonparticipants: Over half of the nonparticipants in
each month were applicants compared with only 35 to 39 percent of the
participants. Several of the other demographic differences between the two
-174- 2 :3 7
TABLE 7.3
SWIM
SELECTEO CHARACTERISTICS OF SWIM-ELIGIBLES AT THE TIME OFINITIAL SWIM REGISTRATION, BY PARTICIPATION STATUS IN
JULY ANO NOVEMBER, 1986
Characteristic
July 1986 November 1986
ParticipantsNon-
Participants
I
ParticipantsNon-
Participants
Office (%)Service CenterSon Diego West
53.0
47.0
52.1
47.9
51.4
48.6
51.3
48.7
AFOC Status (%)Applicant 35.3 50.3* 38.8 54.6***Renewed Recipient 39.7 25.6*** 31.3 21.1***Redetermined Recipient 25.0 24.1 29.9 24.3**
Welfare Status (%)AFOC 66.8 66.4 70.1 63.9AFOC-U 33.2 33.6 29.9 36.1
Average Age (Years) 34.3 33.0*** 34.0 32.9***
Sex (s)Mole 34.7 37.0 31.3 38.7***Female 65.3 63.0 68.7 61.3***
Ethnicity (%)White, Non-Hisponic 24.1 21.4 24.0 23.4
Block, Non - Hisponic 34.5 38.6 34.4 38.1
Hispanic 32.5 35.8 32.0 34.0American Indian/Alaskan Native 0.0 0.10 0.1 0.30Asian and Pacific Islander 7.8 3.5 )** 9.0 3.3111.1,
Other 1.0 0.5 0.4 0.8a
Degree Received (%)High SCh7OI Diploma 45.4 42.0 45.1 44.3GED 7.8 7.2 9.1 6.3 *
None 46.8 50.7 45.8 49.4
M tai Status (%)dyer Married 21.8 27.3** 24.1 28.2
Married, Living wish Spouse 32.2 30.2 29.1 33.8Married, Not Livingwith Spouse 17.7 21.7* 19.9 17.1
Widowed or Divorced 28.3 20.8*** 26.9 20.8***
(continued)
-175- 2.3 8
TABLE 7.3 (continued)
Characteristic
July 1 986
Participants
Non -
Parti ci pants
Mandatory AFDC with ChildLess Than 6 (%)
Monolingual In a Language Other
Than English (%)
SpanishOther
Undocumented Worker (%)
Prior AFDC Dependency (%)Never on AFDC1-11 Months12-23 Months
24-35 Months36-47 Months48-59 Months60 Months or More
4.4
10.8
0.9
0.0
13.9
7.1
8.7
9.4
8.4
7.7
44.8
Average Number of MonthsEver on AFDC 60.8
Average Number of Monthson AFDC During 24 MonthsPrior to Initial Registration
Length of Time Employed During24 Months Prior to InitialRegistration (%)
Not Employed1 Week to 6 Months7-12 Months13-18 Months19-24 Months
Held a Job of Any Time DuringFour Quarters Prtor to initial
Registration (%)
15.6
44.7
17.9
14.2
7.6
15.7
43.4
Estimated Earnings During24 Months Prior to Initial
Registration (%)$0 44.7
C-51(1(10 12.9
$1001-$5000 18.7
$5001-$10,000 12.0
Over $10,000 11.7
2.5*
24.4***7.4
8.1
8.6
6.5
5.6
39.4*
53.5**
12.9***
40.6
20.4
14.6
9.0
15.4
42.7
40.6
17.2**16.4
11.9
13.8
November 1986
Participants
r
Non-
Participants
7.5 3.54gss,
10.4 13.4*
0.6 0.4a
0.0 7.0***
15.0 22.8***
7.9 10.1
7.7 8.4
8.9 6.9
8.7 6.0*
6.4 5.6
45.6 40.2**
57.5 52.9
15.1 12.2***
43.2 39.1
17.3 19.2
12.5 14.3
9.7 9.5
17.3 17.9
44.1 43.4
43.1 39.0
12.3 14.2
19.1 18.2
12.4 12.9
13.1 15.8
(continued)
-176- 239
TABLE 7.3 (continued)
Characteristic
Jul y 1 986 November 1986
ParticipantsNon-
Participants ParticipantsNon-
Participants
Average Earnings During FourQuorters Prior to InitialRegistration ($)
b
Received Unemployment Compensa-tion During 12 Months Priorto Initial Registration (%)
Averoge Amount of UnemploymentCompensation During 12Months Prior to 11nitiolRegistration ($)
Registered Prior toReview Month (%)
2334.52
9.9
143.28
94.6
2127.66
Somple Sizec 1746
9.6
173.90
76.9***
1427
2290.75
10.0
182.41
96.0
2498.70
10.9
161.89
88.9***
1800 1668
SOURCE: MDRC Client Information Sheets, the Stote of Colifornio UnemploymentInsurance earnings ond benefits records, ond the County of Son Diego Deportment ofSocial Services SWIM Automoted Tracking System ond EWEP ottendonce logs.
NOTES: The somple for this toble includes individuols who were eligible forSWIM in July or November, 1986.
The somple is weighted to reflect the actual number of SWIM-Eligibles.
Distributions moy not odd to 100.0 percent due to rounding.
A chi-square or t-test wos applied to differences between participantsond non-participants within eoch review month. Statistical significance levels oreindicated os: = 10 percent; ** = 5 percent; *** = 1 percent.
°Chi - squore test inappropriate due to low expected cell frequencies.
bThese doto ore colculated from the Stote of Col if ornio Unemployment
Insuronce earnings records and Include zero values for somple members not employedond for those not receiving Unemployment Compensation. Individuals without socialsecurity numbers were excluded from these calculations.
cFor selected characteristics, sample sizes may vary up to 5 somple
points due to missing doto.
240177
groups reflect the dominance of applicants among the nonparticipants.
Nonparticipants were more likely to have never been on welfare prior to
registering with the program. Nonparticipants, on average, had less
extensive welfare histories during the two years prior to initial program
registration as well as throughout their life. Nonparticipants were more
likely to have just registered with the program, that is, during either of
the two months; nonparticipants were younger than participants on average.
Finally, nonparticipants were more likely never to have married and less
likely to have been widowed or divorced.
The results highlight two other key differences as well. Between 6
and 7 percent of the nonparticipants were undocumented workers. And
nonparticipants were much less likely than participants to be mandatory
AFDC parents with a child younger than six years old ('soft mandatories").
This is simply a function of WIN registration requirements. Single parents
who are in school are designated az; WIN-mandatory registrants, even though
they have young children. This is because the fact that the parents are in
school is taken to mean that they do not have to take care of their
children Nonparticipants were also less likely than participants to be
Asian.
The results suggest two possible primary reasons for nonparticipation.
Some applicants may have not participated in the program because they were
eligible for program services for only a short time during the month,
perhaps because their applications were never approved. Other applicants,
who might have recently registered with the program, may have been assigned
to activities scheduled to begin the following month.
-178- 2
C. Possible Reasons for Nonparticipation in Selected Months
To analyze reasons for nonparticipation, MDRC staff reviewed the
program case files of a random sample of individuals who, although eligible
for SWIM in July 1986 or November 1986, did not participate during these
months.12 A total of 99 registrants were included in this analysis -- 57
were nonparticipants in July 1986 and 42 were nonparticipants in November
1986.13 The findings suggest that the monthly participation rates observed
in SWIM were close to the maximum feasible in the program.
Most (72 percent) of the inactive registrants were not scheduled
(i.e., expected) to participate during July or November 10,1g, (See Table
7.4.) Only 17 percent were assigned to an activity that was to start
during one of these months or that was scheduled to continue into one of
these months. For 11 percent of the registrants, their case files did not
Indicate whether program staff expected them co participate during July or
November.
Amcng those inactive during July or November 1986, regardless of
whether they welt expected to participate during these months, over 25
percent of the nonparticipants had legitimate !personal' or situational
reasons for not ,zticipating during the reviewed month Included here
were registrants who were ill during the month or caring for someone else
who was ill, and individuals with childcare problems. Very few had
,'personal' reasons for their lack of activity which were considered not
legitimate by the program. Only 5 percent were inactive due to
noncooperation.
Thirty-eight percent of all nonparticipants did not participate during
the review months due to "program' reasons. Over two - fifths of these
-179-242
0301
TABLE 7.4
SWIM
PERCENTAGE DISTRIBUTION OF REGISTRANTS' REASONS FOR NON-PARTICIPATION.BY EXPECTED PARTICIPATION STATUS
Reasons farNon-Participation°
Participation
Expected
ParticipationNot Expected
Expectation forParticipationNot FoundIn Cosefile
Total
Non - Participant
Sample
Non-Participation Du J Registrant's
Personal Situation 64.7 19.7 0.0 25.3***
Registrant's Health Problem 5.9 5.6 0.0 5.1
Other's Health Problem 11.8 2.8 0.0 4.0
Registrant Pregnant 0.0 5.6 0.0 4.0
Child Core Problem 17.6 1.4 0.0 4.0
Housing Problem 0.0 1.4 0.0 1.0
Registrant Moved 5.9 1.4 0.0 2.0
Registrant Uncooperative 23.5 1.4 0.0 5.1
Hon-Participation Due to Soff Delaysor Poor Follow-Up 0.0 18.3 9.1 14.1
Non - Participation Due to Program
Model or Regulations 11.8 50.7 0.0 38.4***
Undocumented Worker 0.0 11.3 0.0 8.1
Less Than 18 Years of Age 0.0 1.4 0.0 1.0
Pending Appeal to Sanction 0.0 1.4 0.0 1.0
Pending Deregistrotion 5.9 5.6 0.0 5.1
Assigned, Woiting for Activity toBegin, SWIM-Eligible Throughout Month 0.0 14.1 0.0 10.1
Assigned, Waiting for Activity toBegin, Registered During Month 0.0 3.5 U.0 6.1
Active In Assessment 0.0 1.4 0.0 1.0
Normal Paperwork Delay 5.9 7.0 0.0 6.1
(continued)
244
TABLE 7.4 (continued)
Reasons forNon-Participation°
Participation
ExpectedParticipationNot Expected
Expectation forParticipationNot FoundIn Coseflle
Total
Non-ParticipantSample
Non-Participation Due to Other Reasonsb
5.9 4.2 0.0 4.0
Reason for Non-Participation Not FoundIn Cosefile
b5.9 0.0 0.0 1.0*
Actually Participated, Incorrect TrackingSystem Data, or Incomplete Cosefile 11.8 7.0 90.9 17.2***
Total 100.0 100.0 100.0 100.0
Total Number of Registrants 17 71 11 99
1-03 SOURCE: MORC calculations from cosefile reviews of randomly selected registrants who were inactive during
July or November 1986.7
NOTES: Distributions may not odd to 100.0 percent and subcategory percentages may not add to mayorcategory percentages due to rounding.
Expected participation status Is defined as whether or not the registrant was 'Tssigned to, or wasexpected to continue In, an activity during the review month.
A chi-square test was applied to differences between the expected participation statuses for majorcategories. Statistical significance levels are indicated as: = 10 percent; ** = S percent; * = 1 percent.
aOnly primary reasons for non-participation are Included In the table. Seven 1 4 had additional
reasons for non-participation including: attended another appointment, assigned (waiting 'er o-tivIty to begin),participated in assessment interview (this does not count as participation), normal paperwork delays, and did notattend assigned activity for some other reason.
b'Other Reasons" for non-participation Include: registrant in loll during the month, change in
registration status, deferred from program to complete public service work, did not attend reappraisal meetingfor good cause. The registrant for whom no reason was found missed a workshop and was automatically rescheduledwith no documented reason.
245 248
individuals had been assigned to a program component and were waiting for
it to begin. Over one-fifth of this group were undocumented workers.
Another group of these individuals were pending deregistrations.
About 14 percent of the nonparticipants w(:e apparently 'lost.' in the
program, reflecting long paperwork delays or a lack of follow-up.
About 17 percent of all sampled nonparticipants actually participated
during th' review r,nths, according to the case files, or their case files
did not contain sufficient documentation to confirm nonparticipation during
the month. Most of these individuals reflected lack of data entry into the
automated tracking system used to select the sample.14
The results indicate some differences in the reasons for nonpart_cipa-
tion between AFDC and AFDC-U registrants, although the small samples neces-
sitate caution in generalizing from these results. AFDC registrants were
more likely, compared to AFDC-U registrants, to be nonparticipants due to
personal situations. Notable here, however, was that very few (5 percent)
of the AFDC nonparticipants were not active due to childcare problems.
Also evident was that AFDC-U's were more likely to be nonparticipants due
to the program model or program regulations. Most notable here was that 24
percent of the AFDC-U nonparticipants were undocumented workers, as
compared to 3 percent of the AFDC nonparticipants.
Further investigation of the individuals in this sampl indicated that
the demographics of the nonparticipants did not differ according to the
type of reason (personal, program modul-relatd, or lack of follow -up) for
inactivity.
A final point, which is very imwrtant, is that a review of the track-
ing data for these 99 individuals indicated that for most of the sample,
-182-
--247
July and November 1986 represented only temporary inactivity. Two-thirds
of the sample participated either before their inactive month or after it.
Only one-third of those who were not participating during the review months
never participated in SWIM.
In unnmary, the monthly participation rates observed in SWIM were
close to the maximum feasible in this program, even though the rates fell
short of the 75 percent goal. First, according to case file reviews, close
to 90 percent of those eligible for program services in any month were
either active or otherwise complied with program requirements during the
month, even if they did not participate. Inactive registrants who were
complying had legitimate personal reasons for not participating; were not
appropriate for assignment to a component; or were assigned to a component
scheduled to begin at a later date. Only one-tenth of those eligible in
any month (about one-fifth cf those inactive during the selected months)
were inactive due to non - cooperation or prograA staff failure to assign or
follow up registrants. Second, the majority of those not participating in
a given month were temporarily inactive; most of these individuals had
participated prior to their 'inactive' month or they participated folic ing
their 'inactive' month.
To corroborate these findings, Client Information Sh (CIS) data,
automated program activity data and welfare data for all those inactive
durir July 1986 and November 1986 were reviewed. These data are not as
detailed ..4s the information found in the case files for the small sample of
registrants. However, they confirm that monthly participation rates of
approximately 50 percent in any giver month probably represent the maximum
that the program could have achieved.
-J 246
The percentages of nonparticipants in either month who fit certain
program situations according to available data are shown in Table 7.5.
These situations were sequentially defined so that each registrant would
fall into only one category. For example, an undocumented worker who was
also registered for less than ten days during the month would be categor-
ized only as an undocumented worker. For comparison, data are also
presented for those who participated during the two months.
Of the July 1986 nonparticipants, 6 percent were undocumented workers
who did not qualify for program services. Another 3 percent were individ-
uals who were ac ive during June 1986 as well as August 1986. These
registrants were probably 'in between' program campon.nts or were temporar-
ily excused from participation, perhaps due to illness. In addition to the
above-mentioned individuals, a large portion of those inactive (28 percent)
participated in either June or August 1986. Again, as of July 1986, these
registrams were experiencing a temporary period of inactivity: They were
'excused' for a short period of had just completed an activity or
were about to start an activity.
Another 8 percent of the nonparticipants were registered for fewer
than ten days during July 1986. These registrants registered and/or
deregistered during the month. It is likely that these individuals did not
participate because they were SWIM-eligible for only a short period of time
during the month. Another 12 percent of the nonparticipants did not
receive welfare during July 1986. Sane of these individuals may have been
applicants whose welfare applications were eventually denied and who, thus,
did not participate. Finally, 26 percent of the July 1986 nonparticipants
were active at some point in SWIM, although they did not participate during
-184-
24
.
iiiIM1=10.1.1M
TABLE 7.5
SWIM
PERCENTAGE DISTRIBUTION OF SWIM-ELIGIBLES, 8Y PARTICIPATION STATUSAND RELATED ACTIVITIES OR STATUSES IN JULY AND NOVEMBER, 1986
Related Activity orStatus
July 1986 November 1986
ParticipantsNon-
Participants ParticipantsNon-
Participants
Undocumented Worker 0.0 6.0*** 0.0 7.0sss
Participated During Month Priorand Month After Review Month 62.7 2.941.1141 72.5 3.10,41
Participated During Month Prioror Month After Review Month 30.6 27.8 23.5 18.6**
Registered Less than 10 DoysDuring Review Month 0.4 Liss* 0.3 4.2***
No AFDC Received DuringReview Month 0.6 12.3*** 0.2 12.9***
Ever Participated (but Non-Participant During Month Frio(and Month after Review Month) 5.7 26.1*** 3.5 35.4***
None of the Above Statuses 0.0 16.9*** 0.0 18.8*"
Total 100.0 100.0 100.0 100.0
Sample Size 1746 1427 1800 1668
SOURCE: AMC calculations from the County of San Diego Department of SocialServices SWIM Automated Tracking System, EWEP attendance logs, and the County of SanDiego AFDC records.
NOTES: The sample for this table Includes individuals who were registered forSWIM In July or November, 1986.
The sample is weighted to reflect the actual number of SWIM-Eligibles.
Distributions may not odd to 100.0 percent due to rounding.
Participation Is defined as attending EWEP for at least one hour or anyother activity for of least one day.
A chi-square test was applied to differences between participants andnon-participants within each review month. Statistical significance levels areindicated as: * = 10 percent; ** = 5 percent; *** = 1 percent.
250-185-
the summer of 1986.
The 17 percent of the July 1986 nonparticipants who did not meet any
of the above criteria were probably in a variety of situations. As indi-
cated by the results of the case file reviews, some of these individuals
may have been lost! in the program; a number of them may have had
deregistrations pending; and some may have been uncooperative and
subsequently sanctioned.
The distribution of nonparticipants according to these characteristics
was similar in the two months examined. Again, the results indicate that
monthly participation rates of approximately 50 percent in any given month
probably represent the maximum feasible rates that this program could have
achieved.
*
As shown in this chapter, several different kinds of monthly
participation rates can be calculated. Table 7.6 summarizes various rates
for two months of the demonstration -- July and November 1986. The table
illustrates many of the points made in the cb:pter.
Varying the types of activities which count as participation had the
biggest effect on monthly participation rates. Varying the definition of
program-eligibles, and varying the time period over which the rates are
calculated, each had a small effect on the rates. Taking into account the
participation of registrants prior to each month as well as subsequent to
each month also had a substantial effect on the rates: Many of those
inactive during specific months were only temporarily inactive. Lastly,
-186-
TABLE 7.6
SWIM
ALTERNATIVE ACTIVITY MEASURES FOR JULY AND NOVEMBER 1986
Definition of Rote July 1986 November 1986
Varying Type of Activity Counted asParticipation
Percent of SWIM-Eligibles Registered of LeastOne Doy Duri_a Month Who, During the Month,Porticipated in:
Progrom-Arranged Activities
Progrom-Arranged Activities or Self-InitiatedEducation or Training
Progrom-Arranged ActPeities, Self-lnitiotedEducotion or Troining, or EmploymentWhile Registered
Varying Definition of Program-Eligibles
Percent of SWIM-Eligibles Who Participated inProgram-Arronged Activities, Self-InitiatedEducotion or Troinino, or Employment WhileRegistered During the Month, Where 'SWIM-Eligibles' ore Defined as Those:
Registered Throughout the MonthRegistered as of the End of the MonthRegistered at Leost One Day in the Month
Varying Time Period Over Which Ratesare Calculated
Percent of Individuals Registered of LeostOne Day During the Third or Fourth Quorter of1986 Who Participated in Program-ArrangedActivities, Self - Initiated Education orTraining, or Employment While RegisteredDuring the Quorter
Taking Into Account Temporary Periodsof Inactivity
Percent of SWIM- Eligibles Registered at LeastOne Day During the Month Who Participated in
Program-Arranged Activities, Self-InitiatedEducotion or Training, or Employment WhileRegistered During the Month or:
in the Prior Month as well as theFollowing Month
In the Prior Month or the Following Month
28.0%
35.4
55.0
62.7
56.9
55.0
60.5
21.3%
32.8
51.9
55.4
52.9
51.9
57.6
56.3 53.468.8 62.4
-187- 252
(continued)
TABLE 7.6 (continued)
SOURCE: MDRC calculations from the County of Son Diego Department of SocialServices SWIM Automated Trucking System and EWEP attendance logs.
NOTES: Participation is defined as attending EWEP for at least one hour orany other activity for at least one day. Program - arranged activities include JobSearch Workshop, EWEP, Job Club, STAR, ISESA, OJT, or any program-arranged educationor training. Solf-initiated activities include education, training, union, or otherjob search activities. Only employment that occurred while on individual wasregistered is included.
253
-188-
and not shown in the table, the monthly participation rates achieved in
SWIM appear to be close to the maximum feasible rates this program could
have achieved: Most registrants who were inactive during specific months
were complying with program requirements during the month, even if they
were not participating. Only a small pronortion (about 10 percent) of
registrants eligible for services during a month were not active due to
noncooperation or program staff failure to assign or follow-up appropriate
registrants.
25 4-189-
CHAPTERS
IMPACTS ON EMPLOYMENT, EARNINGS AND WELFARE OUTCOMES
This chapter presents preliminary short -term impacts of the SWIM
program on employment, earnings, welfare receipt and welfare payments. The
findings are presented first for AFDC registrants, who are primarily heads
of single-parent families (and female), and then for AFDC-U registrants,
who are adults in two- parent families (and mostly male).
For AFDC registrants, SWIM achieved statistically significant gains in
employment and earnings, as well as reductions in the percentage receiving
welfare and in welfare payments. For AFDC-U registrants, SWIM resulted in
statistically significant gains in employment and reductions tn welfare
payments. However, increases in earnings were not always statistically
significant.
I. Analysis Issues
This chapter addresses two questions: What were the employment,
earnings and welfare outcomes of those enrolled in the program? And what
would these outcomes have been had the program not existed? These
questions are answered by examining the behavior of an experimental group,
eligible for SWIM services, as compared to a control group, which was
similar in all respects but not offered SWIM services. Control group
members were free to engage in self-initiated education and training
activities but were not provided program services. The differences between
the average outcomes for the experimental group and the control group yield
-190-55
estimates of program impacts. Outcome differences between experimentals
and controls were considered statistically significant if there was no more
than a 10 percent possibility they could have occurred by chance.
To ensure that the two groups were the same in measured and unmeasured
characteristics, as noted in Chapter 2, AFDC and AFDC-U registrants were
randomly assigned to either the experimental or the control group. The use
of randomization in constructing the control group is the linchpin of
experimental research. Successful randan assignment assures that
motivational and other unobserved differences among individuals will be
balanced between the experimental and control groups in the comparison of
final outcome differences.
The control group provides a benchmark of the normal employment and
welfare turnover behavior. Even wPhout the program, some individuals find
employment or leave welfare within a relatively short period of time.
Experimental-control differences enable this normal turnover to be excluded
fran the impact estimates.
In fact, the normal turnover was substantial. Although most controls
received welfare benefits during the first quarter of follow-up, many found
work and left welfare relatively quickly without special assistance.
Further, the registrants' own earnings are not the only way off welfare.
Among controls, for example, more than half of those who received no
welfare payments during the fourth quarter also had no earnings. As is
known fran other research, registrants may leave welfare because of
marriage or reconciliation, children who age out of dependent status,
increased earnings of other family members, and other changes in family
cirowistaAces. 1
256'-191-
Although there is considerable information about th,J extent of partici-
pation of SWIM experimentals in a wide range of activities, as indicated in
the prior chapters in this report, little is known about the services
received by the control group. Although controls did not participate in
the SWIM program, at registration 15 percent of the AFDC controls and 10
percent of thr AFDC-U controls were in sane type of education and training
programs. Controls, on tveir initiative, may have also entered education
and training programs at any time during the follow-up period. (The final
report will provide more information on the activities of the control
group.)
It is important to recognize, however, that randomization dictates the
comparisons that can be made and, therefore, the impacts that can be
estimated. First, all persons randomly assigned must be included in the
impact calculations in order for the estimates to be unbiased. Since
nonparticipants were reached by sanctioning, and sanct..oning can result in
impacts, particularly on welfare payments, any effect of sanctions on
nonparticipants must be included in program impact estimates. Thus,
comparisons of all controls with all experimentals, including nonparti-
cipants and participants, are appropriate in evaluating a mandatory program
such as SWIM.
Second, earnings and AFDC payment estimates must also include sample
members not enployed or not receiving welfare, assigning them zero dollar
values. To the extent that the program converts non-earners into earners,
or welfare recipients into non - recipients, exclJsion of zero values from
both the control and experimental group estimates could lead to an under-
estimate of program impacts.
Unemployment Insurance earnings data were collected by calendar
quarter. Since random assignment could have taken place at any point
during this three-month period, the quarter in which random assignment
occurs may include earnings before random assignment. The AFDC monthly
payments data were aggregated into calendar quarters in order to match
these earnings measures. Therefore, the quarter of randcw assignment is
not counted as follow-up quarter for cumulative impact estimates.
In the present analysis, data are available for examining impacts for
a relatively short period after random assignment. The employment and earn-
ings data are available for three, and the MDC data for four, additional
quarters beyond the quarter of random assignment. The final report will
include a longer period of follow-up and provide a better assessment of
longer-term impacts.
II. Impacts for the AFDC Registrants
This section examines the impacts of SWIM on employment, earnings,
welfare outcomes and measured income (defined as the sum of welfare and
earnings) for AFDC registrants. It also looks at differences in impacts
between applicants and recipients, and discusses preliminary evidence on
longer-term impacts.
A. Employment and Earnings Impacts
Table 8.1 and Figure 8.1 present impacts of the SKIM program among the
AFDC group during the three-quarter follow-up period. 2 As noted, SWIM had
statists 'ally significant impacts for AFDC registrants on all four major
outcomes: increased employment, higher earnings, less time on welfare and
reduced welfare payments. Over the time period as a whole, experimentals
258-193-
TABLE 8.1
SWIM
ALL AFDC: SHORT -TERM IMPACTS ON EMPLOYMENT,
EARNINGS, WELFARE RECEIPT, AND WELFARE PAYMENTS
Outcome and Follow-Up Period Experimentals Controls Difference
Ever Employed, Quarters 2-4 DO
Average Number of Quarters with
46.4 36.4 +10.0***
Employment, Quarters 2-4 0.97 0 76 +0.22***
Ever Employed (%)
Quorter of Random Assignment 27.9 25.1 +2.70*Quarter 2 30.8 24.6 +6.3***Quarter 3 32.9 25.3 +7.6***Quarter 4 33.5 25.7 +7.8***
Average Total Earnings,Quarters 2-4 ($) 1442.00 1185.47 +256.54***
Average Total Earnings ($)Quarter of Random Assignment 296.68 285.23 +11.44Quarter 2 371.00 338.25 +32.76Quarter 3 497.69 392.03 +105.66***Quarter 4 573.31 455.19 +118.12***
Ever Received Any AFDC Payments,Quarters 2-5 (%) 91.1 91.9 -0.7
Average Number of Months ReceivingAFDC Payments, Quarters 2-5 8.59 9.12 -0.53***
Ever Received Any AFDC Paym6Ats (%)Quarter of Random Assiroment 91.2 91.4 -0.3Quarter 2 89.7 89.7 -0.1Quarter 3 78,9 81.5 -2.6**Quarter 4 70.6 76.0 _5.5::*
Quarter 5 65.8 72.4 -6.5***
Average Total AFDC PaymentsReceived, Quarters 2-5 ($) 4424.00 4827.08 -403.08***
Average AFDC Payments Received ($)Quarter of Random Assignment 1193.27 1194.12 -0.86Quorter 2 1286.17 1331.93 -45.76**
Quarter 3 1119.45 1224.55 -105.10***Quarter 4 1031.55 1159.81 -128.27***Quarter 5 986.83 1110.78 -123.95***
Sample Size 1606 1605 3211
-194- 25a
(continued)
TABLE 8.1 (continued)
SOURCE: MAC calculations from the County of Son Diego AFDC records andthe State of California Unemployment Insurance earnings records.
NOTES: The sample for this tcbie includes individuals who registeredbetween July 1985 and June 1986.
These data Include zero values for maple members nct employed andfor sample members not receiving welfare. these data ore regression-adjustedusing ordinary least squares, controlling or pre-random assignmentcharacterlstics of sample members. There may be some discrepancies incalculating sums and differences due to rounding.
For all moosure'., the quarter of random assignment refers to theMender quarter in which random assignment occurred. Because Quarter I. the
quarter of random assignment. may contain some earnings and AFDC payments fromthe period prior to random assignment. it is excluded from the summary measuresof follow-up.
A two-tolled t-test was applied to differences betweenexperimental one ontrol groups. Statistical significance levels are indicatedas: * = 10 perct t; ** = 5 percent; *** = 1 percent.
195 260
50 ;-
;
0
FIGURE 8.1
SWIM
AFDC SHORT-TERM IMPACTS ONEMPLOYMENT AND EARNINGS
......... ., ...........- ....
prior Quarter 2 QuartQuaer 1prior der of Rend°In Quarter 2 Quarter 3 Quartet 4
Qua
600
400400 I-
200
0
Quarter Relative to Random Assignment
pro( Quarter 2 Quartsr(Iuarter ot Random
AssignmentQuarter 2 Quartet 3
Quarter Relative to Random Assignment
196
Qualto; 4
Experimentals
Controls
Experinentals
Controls
were more likely to be employed than controls: 46.4 percent of experiment-
als ir, contrast to 36.4 percent of controls, a statistically significant
difference of 10 percentage points. Employment gains continued over the
three follow-up quarters. In quarter 2, the employment rate for experi-
mentals was 6.3 percentage points higher than the control average employ-
ment rate of 24.6 percent. By the fourth quarter, the employment rate
among experimentals was 7.8 percentage points higher than the control
average employment rate of 25.7 percent.
Experimentals also earned $257 more than the control group average of
$1,185 over the follow-up period as a whole -- a statistically significant
increase of 22 percent. Statistically significant earnings increases were
found beginning in the third quarter. By the final quarter, earnings gains
peaked at $118.
Two different factors could have accounted for this earnings increase:
(1) an increase in employment rates and/or (2) higher earnings (either more
hours worked or higher average hourly wage rate or sane combination) among
those employed. A comparison of the earnings of employed experimentals and
employed controls (presented in Appendix Table F.2) suggests that the earn-
ings impacts are primarily due to increased employment. In fact, earnings
among those employed werl slightly lower for experimentals ($1,711 in
quarter 4) than controls ($1,772 in quarter 4). Thus, the effect of the
program seeas to be to encourage those who would not have worked in the
absence of the program to work, at least in the short run, rather than
creating opportuni ies for higher earnings among those who would have
worke,1 without the program. However, it should be noted that this is a
non-experimental comparison; the characteristics of experimentals who are
-197- 262
employed may be different from those of employed controls.
B. Welfare Impact
The impacts of the SWIM program on the percentage receiving welfare
and welfare payments are shown in Table 8.1 and Figure 8.2. As expected,
SWIM did not affect the percentage of registrants who received welfare
during the full follow -up period, since the proportion of both experi-
mer 31s and controls receiving welfare during the quarter of random
assignment is similar for both groups. On average, experimentals were
receiving welfare benefits for fewer months than controls. By the end of
the four-quarter follow-up period, there were statistically significant
reductions in welfare receipt. In the last quarter of follow-up, for
instance, the proportion of experimentals receiving welfare was
approximately 6.5 percentage points less than the control group rate of
72.4 percent.
There were statistically significant welfare savings as well, both
during the full follow-up period as well as quarter by quarter. Expert-
mentals received $403 less in welfare payments during the four- quarter
follow-up period than the control group average payment of $4,827. Welfare
savings increased through the fourth quarter and then leveled off. By the
fifth quarter, welfare payments to experimentals were $987 as compel to
$1,111 for controls, yielding a grant reduction of $124.
C. Other Impacts
Table 8.2 presents program impacts on the distribution of earnings,
the mixing of earnings and welfare, and measured income (which is the sum
of earnings and welfare payments). The fourth quarter is used for this
analysis because it is probably most indicative of the longer-term impacts.
-198-
2 3
100r
50 r
0
FIGURE 8 . 2
SWIM
AFDC: SHORT-TERM IMPACTS ONAFDC RECEIPT AND PAYMENTS
prior Wader 21,60c wetter 1
a"gervelorn wartet 2 Queer 3 Wester 4 Quarter 5
Qu Ass'
1500 -
1400 ;-
1300
1200
1100
1000.....
900
800
700
600
500 '2 der 1
Rio( Queer prior Qua water ot gand°111,,,4
p,s60111'''
Quarter Relative to Random Assignment
Wader 2 Quarter 3
Quarter Relative to Handom Assignment
-199-264
199-
....... .........
Quartet 4 Quartor 5
2 64
Experinentals
Controls
Ex, Amentals
Controls
TABLE 8.2
SWIM
ALL AFDC: SHORT -TERM IMPACTS CN DISTRIBUTION OF EARNINGS,EARNINGS/WELFARE MIX, AND ME&SURED INCOME IN QUARTER FOUR
Employment and Welfare Outcomes Experimentals i.ontrols Difference
Average 'iota; Earnings.Quarter 4 (%)
None
$1 - $1500
More Than $1500
Total
66.5
18.1
15.5
100.0
74.3
13.7
12.1
100.0
_7.3***
+4.4**-/
+3.44**
Employment and Welfare Status,Quarter 4 (%)
Had No Earnings, Received : meAFDC Payments 50.4 60.4 _9.9***
Haa No Earnings, Received NoAFDC Payments 16.0 13.8 +2.1*
Had Some Earnings, Received SomeAFDC Payments 20.1 15.7 +4.4***
Had Some Earnings, aceived NoAFDC Payments 13.5 10.1 +3.ess*
Total 100.0 100.0
Average Measured Income,Quarter 4 (%)a
None 16.0 13.8 +2.1*
$1 - $1500 35.2 37.6 -2.4
More Than $1500 48.9 48.5 +0.3
Total 100.0 100.0
Average .measured Income ($)a 1604.86 1615.00 -10.15
Sample Size 1606 1605 3/11
-200-
(continued)
TABLE 8.2 (continued)
SOURCE: MORC calculations from the County of Son Diego AFDC records andthe State of California Unemployment Insurance earnings records.
NOTES: The sample f this table includes individuals who registeredbetween July 1985 onC June 1986.
These data include zero values for sample members not employedand for sample members not receiving welfare. These d6to ore regression-adjusted using ordinary least squares, controlling for pre-rondom assignmentchoracteristics of somple members. There may be some discrepancies incalculating sums and differences due to rounding.
For ull measures, the quarter of random assignment refers to thecalendar quarter in wnlch random assignment occurred. Because Quarter 1, thequarter of random assignment, may contain some earnings and AFDC payments fromthe period prior to random assignment, it is excluded from the summary measuresof folio -up.
A two-tailed t-test wns applied to differences betweenexperimental and control groups. Statistical significance levels ore indicatedas: * = 10 percent; ** = ''. percent; *** = 1 percent.
oAveroge measured income is defined os personal earnings plus
welfare payments received during a quarter.
201I
The first panel of Table 8.2 shows the impact of SWIM on the earnings
distribution. SWIM moved experimentals out of the no earnings category
into both the category with quarterly earnings of $1-$1,5r, and the cate-
gory with quaiLerly earnings over $1,500. Experimentals were less likely
to have no earnings in the fourth quarter than controls, a statistically
significant dif.!erence of 7.8 percentage points. This, combined with the
fact that the experimental- control differences in both earnings categories
were similar, also supports the notion that earnings gains were due to
increases in employment rather than to higher earnings for those employed.
The second panel shows the impact of SWIM on AFDC registrants' com-
bined income and welfare. By quarter 4, SWI had reduced the proportion of
registrants receiving welfare and no earnings, and increased the proportion
who received both welfare and earnings and the proportion who were
completely off welfare and receiving earnings. There was also a small, but
statistically significant, increase in the proportion receiving neither
welfare nor earnings. These individuals may be substituting income from
other family members' they may also have decided to leave welfare rather
than participate in a mandatory program.
The bottom panel of Table 8.2 presents the impacts on measured income,
which is the sum of the individual's own earnings plus the amount of
welfare payments recorded for that case. There were no statistically
significant differences in average measured income between experimentals
and controls in any follow-up quarter. (See Appendix Table F.3.) Further,
experimentals were slightly more likely than conz:rols tc have no measured
incase. These findings indicate that earnings gains were offset by
reductions in AFDC rayments for the AFDC sample as a whole.
-202-
D. Impacts among Applicant And Recipients
Evaluations of other programs included in MDRC's Demonstration of
State Work/Welfare Initiatives have revealed differences in impacts for
persons applying :.or welfare at the point they were randomly assigned
(called applicants) versus individuals already on the welfare rolls at
random assignment (called recipients). Table 8.3 presents impacts
separately for the two groups. Approximately 40 percent of the total AFDC
sample are applicants and 60 percent are recipients. As documented in
Chapter 2, applicants generally are less disadvantaged and more job-ready
than recipients, as measured by their employment and earnings history as
well as prior welfare dependency. This is further supported by the higher
employment rates and earnings and lower welfare receipt among applicant as
compared to recipient controls during the follow-up period.
As indicated in Table 8.3, SWIM resulted in statistically significant
gains in employment and earnings and reductions in welfare receipt and
payments among both AFDC applicants and recipients.
Among AFDC applicants, 52 percent of the experimentals were employed
at sane point during the three-quarter follow-up period as compared to 44.4
percent of the controls. Thi: resulted in a statistically significant
increase of 7.6 percentage points. Quarterly employment gains increased
over the follow-up period. Applicant experimentals earned $1,882 during
the full follow-up period as compared to the control group average earnings
of $1,545, yielding a statistically significant earnings gain of $338.
Earnings gains also increased after the second quarter.
Among AFDC recipients, 42.7 percent of experimentals and 31.5 percent
of controls were employed at sane point during the full follow-up period,
-203- 263
TABLE 8.3
SWIM
ArOC APPLICANTS ANO RECIPIENTS: SHORT-TERM IMPACTSON EMPLOYMENT. EARNINGS. WELFARE RECEIPT. ANO WELFARE PAYMENTS
Outcome and Follow-Up Period
AFOC Applicants° AFOC Recipientsa
Experimenter's Controls Oifference Experimentals Controls Oifference
Ever Iployed, Quarters 2-4 (s) 52.0 44.4 +7.6*** 42.7 31.5 +11.2***
Average Number c' Quarters withEmployment, Qua ters 2-4 1.07 0.94 +0.13** 0.91 0.65 +G.26***
Ever Employed (%)Quarter of Random Assignment 37./ 34.6 +3.0 21.6 19.1 +2.5*Quarter 2 34.1 31.0 +3.1 28.5 20.8 +7,7***Quarter 3 36.7 32.5 +4.2* 30.4 20.9 +9.5**oQuarter 4 36.5 30.5 +6.0** 31.5 22.8 +8.7***
Average Total Earnings,Quarters 2-4 ($) 1882.14 1544.63 +337.51* 1158.24 952.00 +206.25**
Average Total Earnings ($)Quarter of Random Assignment 458.10 423.99 +34.11 194.68 193.28 +1.40Quarter 2 489.03 446.16 +42.b7 293.02 270.33 +22.69Quarter 3 655.54 537.96 +117.58* 395.98 297.72 +98.26***Quarter 4 737.57 560.51 +177.06** I 469.24 383.94 +85.30*
Ever Received Any AFOC Payments,Quarters 2-5 (%) 85 0 85.2 -0.2 94.9 96.2 -1.3
Average Number of Months ReceivingAFOC Payments, Quarters 25 6.92 7.43 -0.51* 9.66 10.21 -0.55***
Ever Received Any AFOC Poyments (%)Quarter of Random Assignment 81.3 82.7 -1.4 97.3 97.3 -0.1Quarter 2 82.1 81.3 +0.9 94.4 95.4 -1.0Quarter 3 65.0 69.7 -4.7* 87.8 89.2 -1.4Quarter 4 56.1 61.5 -5.4** 79.8 85.5 -5.6***Quarter 5 51.4 57.8 -6.44* 75.2 81.7 -6.6***
Average Total AFOC PaymentsReceived, Quarters 2-5 ($) 3372.03 3723.32 -351.28** 5095.14 5542.81 -447.67***
Average AFOC Payments Received ($)Quarter of Random Assignment 712.10 702.15 +9.95 1501.24 1512.59 -11.35Quarter 2 1049.72 1096.34 -47.12 1437.00 1485.45 -48.45**Quarter 3 833.16 950.60 - 117.44 * ** 1302.15 1403.04 - 100.89 * **Quarter 4 764.11 867.55 -104.84** 1201.90 1347.59 -145.69***Quarter 5 724.94 806.82 -81.88* 1154.09 1306.73 -152.64***
Sample Size 647 611 1258 959 994 1953
2i".;;) (continued)
TABLE 8.3 (co:M., ..;c1)
SOURCE: AMC calculations from the Countt. of Son Diego AFDC records and the State of California UnemploymentInsurance earnings records.
N5TES: The sample for this table - udes individuals who refiistered between July 1985 and June 1986.
These data include zero values for sample members :lot employed and for sample members not receiving welfare.These data are regression-adjusted using ordinary least squares. controlling for pre - random assignment characteristics ofsample members. There may be sane discrepancies in calculating suns and differences due to rounding.
For all measures. the quarter of random assignment refers to the calendar quarter in which random assignmentoccurred. Because Quarter 1. the quaTter of random assignment. may contain sane earnings and AFDC payments Iran the periodprior to random assignment. It is excluded f ran the sumnory measures of follav-up.
A two-tailed t-test was applied to differences between experimental and control groups. Statisticalsignificance levels are indicated as: *= 10 percent; **= 5 percent; *Qs = 1 percent.
aRegressions were run on separate subsompl es of applicants and recipients.
27u
resulting in a statistically significant 11.2 percentage point increase.
Recipient experimentals earned $1,158 as compared to $952 for controls
during the follow-up period, a statistically significant gain of $206.
Employment and earnings gains were observed throughout much of the
follr-up period.
SWIM resulted in reductions in welfare receipt and payments during the
follow-up period for AFDC applicants. By the fifth quarter, there was a
6.4 percentage point reduction in the propertion of experimentals receiving
welfare, over the control group average of 57.8 percent. While applicant
experimentals received $3,372 in welfare payments zing the full four-
quarter follow-up period, controls averaged $3,723 in payments. This
resulted in welfare savings of $351. Pihile grant reductions for applicants
peaked in quarter 3, they were still present in the final quarter of
follow-up.
Alaong AFDC recipients, SWIM also resulted in a lower proportion on
welfare and lower welfare payments. In the fifth quarter, the proportion
of experimental recipients receiving welfare was 75.2 percent, 6.6 per-
centage points lower than the control group average of 81.7 percent.
Recipient cilxperimentals received $5,095 in welfare payments during the full
follow-up period as compared to $5,543 for controls, which represented a
statistically significant w&.fare savings of $448. Reductions in grant
payments increased throughout the follow-up period.
E. Longer-term Impacts
One important issue is the effect of the SWIM program on outcomes
beyond the fourth or fifth quarter. Preliminary examination of longer-term
findings is based on the employment and welfare behavior of registrants who
-206-
271
entered the sample between July 1985 and December 1985. For these
registrants, two additional quarters of follow-up are available.
One potential issue with using the early-enrolling sample is the
extent to which it is representative of later enrollees and, hence, an
indicator of longer -run impacts for the full sample. The early-enrollee
sample had somewhat different characteristics from later enrollees. As
indicated in Chapter 2, early enrollees were more likely to be applicants,
and less likely to be renewed recipients.3 Although the earlier ple
appears to be less disadvantaged with respect to prior welfare dependency
than7the later one, these differences are not very large. And examination
of the impacts for the first three quarters suggest that early and later
enrollees have similar impacts, at least in the short-run. (See Appendix
Table F.4.)
As indicated in Table 8.4, SWIM resulted in statistically significant
em;:loyment and earnings gains for the early AFDC sample that ware sustained
during the additional quarters of follow-up. By the fifth z.nd final
quarter, the employment rate for experimentals was 8 percentage points
greater than the control group employment rate of 26.8 percent, and the
earnings gains were $149 greater than the control group average earnings of
$510.
SWIM also resulted in reductions in the proportion of experimentals
receiving welfare, accompanied by welfare savings, that continued through
the six quarters of follow-up. By the final quarter of follow-up, the
proportion of experimentals receiving welfare was 6.9 percentage points
less than the control group rate of 62.3. Average welfare payments in this
final quarter were $120 less than the control group average payment of
-207- 2"/
TABLE 8.4
SWIM
AFDC EARLIER COHORT: LONGER-TERM IMPACTS ON EMPLOYMENT,EARNINGS, WELFARE RECEIPT, AND WELFARE PAYMENTS
Outcome and Follow-Up Period Experimentals Controls Difference
Ever Employed, Quorters 2-6 (%) 55.5 45.6 +9.9***
Averoge Number of Quorters withEmployment, Quarters 2-6 1.73 1.37 +0.36***
Ever Employed (s)
Quorter of Rondom Assignment 29.0 26.8 +2.2Quarter 2 32.6 25.6 +7.0***Quarter 3 33.9 27.1 +6.8***Quarter 4 36.5 28.3 +8.2***Quarter 5 5.2 28.7 +6.5***Quarter 6 54.9 26.8 +8.0***
Averoge Total Earnings,Quarters 2-6 ($) 2806.15 2295.67 +510.48 **
4ierage Totai Earnings ($)Quorter of Rondom Assignment 315.99 301.93 +14.05Quorter 2 392.50 350.02 +42.48Quorter 3 496.05 417.12 +78.93*Quorter 4 623.57 491.28 +132.29**Quarter 5 634.53 526.97 +107.56*Quorter 6 659.52 510.28 +149.23***
Ever Received Any AFDC Poiments,Quorters 2-7 (%) 90.7 90.9 -0.2
Averoge Numoer of Manths ReceivingAFDC Payments, Quorters 2-7 11.59 1 ?.54 -0.95***
Ever Received Any AFDC Payments (%)Quorter of Rondom Assignment 91.1 90.1 +1.0Quarter 2 88.6 88.4 +0.1Quarter 3 77.2 79.2 -2.1Quorter 4 69.0 73.3 -4.3**Quarter 5 63.5 70.6 -7.0***Quorter 6 58.8 66.1 ..7.3**10
Quorter 7 55.4 62.3 -6.9***
(continued)
'13
1111111=111=111111110r1111111EW4111611111111=1=111111111M1111111
TABLE 8.4 (continued)
Outcome and Follow-Up Period Experimentols Controls Difference
Average Total AFDC PaymentsReceived, Quarters 2-7 ($)
Average AFDC Payments Received (S)Quarter of Random AssignmentauarIer 2()varier 3
Quarter 4Quarter SQuarter 6Quarter 7
:992.48 6637.17 -644.69***
1197.57 1169.10 +28.481253.29 1298.42 -45.12*1072.21 1168.28 -96.07***988.82 1103.22 -114.40***944.28 1079.58 -135.',0***
890,80 1024.76 -133 *6843.07 962.91 -119.,JS"*
Sample Size 870 888 1758
SOURCE: MDRC calculations from the County of San Diego AFDC records andthe State of California Unemnloyment Insurance earnings records.
NOTES: The sample for this table includes individuals who registeredbetween July and December, 1985.
These data include zero yr:J:2es for saml1le members not employedand for sample members not receiving welfare. These data are regression-adjusted using ordino:y least squares, controlling for pre-random assignmentcharacterisics of sample members. There may be some discrepancies incalculating sums and differences due to rounding.
Far all measures, the quarter of random assignment refers to thecalendar quarter in which random assignment occurred. Because Quarter 1, the
quarter of random assignment, me, contain some earnings and AFDC payments fromthe period prior to random assignment, it is excluded from the summary measuresof follow-up.
A two - tailed t-test was applied to differenceF bc.iween
experimental and control groups. St "tistical significance levels are indicated
as: = 10 percent; 10 = 5 percent; = 1 percent.
2 4
209
$963. These preliminary figures on longer-term impacts for a group of
early-enrollees suggest that SWIM impacts for AFDC's are likely to extend
beyond the follow-up period available for this report.
III. Impacts For the AFDC-U Registrants
This section of the report examines the same set of impacts for AFDC-U
registrants as have just been discussed for the AFDC sample. It should be
noted that the AFDC-U sample is smaller than the AFDC sample. Hence, it
may be less likely that impacts of a similar magnitude will be statis-
tically significant.
A. Employment and Earnings Impacts
SWIM resulted in statistically significant increases in employment
among AFDC-U registrants. (See T-ble 8.5 and Figure 8.3.)4 During the
full three-quarter follow-up period, the proportion of experimentals who
were employed at same point was 9.3 percentage points greater than the
control group average employment rate of 44 percent. While SWIM resulted
in statistically significant employment gains beginning in the third
quarter, there was some decline in these imr-Acts over time.
Gains in earnings were statistically significant during the Zollow-up
period as a whole.5 Earnings among experimentals were $337 greater than
the control average of $2,028. While there were increased earnings for
experimentals beginning in the second quarter, only in the third quarter
were these impacts statistically significant.
B. Welfare Impactr
Although SWIM did not reduce the proportion of AFDC-U registrants
receiving welfare, there were statistically significant reductions in
-210- rt.i,-.
TABLE 8.5
SWIM
ALL AFDC-U: SHORT-TERM IMPACTS ON EMPLOYMENT.EARNINGS. WELFARE RECEIPT, ,AD WELFARE PAYMENTS
Outcome and Follow-Up Period Experimentals Controls Difference
Ever Employed. Quarters 2-4 (%) 53.4 44.0 +9.3***
Average Number of Quarters withEmployment, Quarters 2-4 1.13 0.94 +0.19***
Ever Employed (%)Quarter of Random Assignment 37.9 35.8 +2.1Quarter 2 36.3 29.2 +7.1***Quarter 3 37.9 31.7 +6.2**Quarter » 38.9 32.8 +6.0 $
Average Total Earnings.Quarters 2-4 (S) 2364.77 2027.77 +337.01*
Average Total Earnings ($)Quarter of Random Assignment 600.73 601.'4 -0.71Quarter 2 656.56 564.15 +92.41Quarter 3 837.00 692.33 +144.67*Quarter 4 871.21 771.29 +99.93
Ever Received Any AFDC Payments.Quarters 2-5 (%) 86,2 86.4 -0.3
Average Number of Months ReceivingAFDC Payments, Quarters 2-5 7.57 7.93 -0.37
Ever Received Any AFDC Payments (%)Quarter of RP 'om Assignment 85.6 84.4 +1.7.
Quarter 2 83.4 83.9 -0.5Quarter 3 67.5 71.0 -3.5Quarter 4 64.7 67.4 -2.7Quarter 5 59.9 62.7 -2.8
Average Total AFDC PaymentsReceived, Quarters 2-5 ($) 4873.97 5298.34 -424.37**
Average AFDC Payments Received ($)Quarter of Random Assignment 1263.87 1274.29 -10.42Quarter 2 1418.75 1470.22 -51.48Quarter 3 1191.54 1321.01 -129.47***Quarter 4 1165.63 1279.25 -113.62**Quarter 5 1098.05 1227.85 -129.80**
Sample Size 687 654 1341
SOURCE AND NOTES: See Table 8.1.
100
50
0
FIGURE 8 3
SWIM
AFDCU: SHORTTERM IMPACTS ONEMPLOYMENT AND EARNINGS
..................... ..........
pAor Quarter 2 Prior Quarter Quot Random
a7,ssIgnment
woo r
800
600 r
400
200
0
(Waiter 2 Quarter
Quarter Relative to Random Assignment
.....................
Qatar 4
................................
u 2 Quarter I?de( Q-arter ?TO nexter ot Ralirti Quarter 2 Quarter 3 Quarter 4
AssVanmei""
Quarter Relative to Random Assignment
,:212
Experimentals
Controls
Expel imentals
Controls
welfare payments over the follow-up period as a whole and beginning in
quarter three. (See also Figure 8.4.) During the full four-quarter
follow-up period, experimental AFDC-Us received $4,874 in welfare payments
as compared to $5,29° for controls, yielding a statistically significant
reduction of $424.
C. Other Impacts
Table 8.6 presents the distribution of earnings and income in quarter
4, che last quarter for which there are both earnings and welfare data.
SWIM reduced the proportion of AFDC-U registran_s with no earnings in this
quarter. (See first panel of Table 8.6.) SWIM also reduced the proportion
of AFDC registrants with only welfare income, but increased the proportion
who combined earnings and welfare payments. (See second panel of Table
8.6.)
As indicated in the third panel of Table 8.6 and Appendix Table F.7,
SWIM did not increase measured income. This suggests that, as for the AFDC
group, SWIM's earnings gains were off et by welfare reductions for AFDC-U
registrants, at least in the short-run.
D. Impacts among Applicar_ts and Recipients
Table 8.7 presents impacts separately for AFDC-U applicants and
recipients.6 Approximately 60 percent of the AFDC-U's were applicants and
40 percent were recipients. Similar to AFDC registrants, applicants are
generally more employable and less dependent than recipients. ':his can be
seen from tne higher levels of employment and earnings, and lower receipt
of welfare and average welfare payments, among applicant controls than
recipient controls.
SWIM resulted in statistically significant employment gains for both
-213-2:/a
100
50
FIGURE 8 4
SWIM
AFDC-U: SHORT-TERM IMPACTS ONAFDC RECEIPT AND PAYMENTS
0
.... ......
t 2Prior Cttlec prior Quartet ot tiand°1
Qua Assignment
ft
as
1500
1400
1300
1200
1100
1000
900
800
700
600
(Natter 2 Quartet 3
Quarter Relative to Random Assignment
Quarter 4 Quarter 5
Experinentals
Controls
Experimentals
;-!
Controls
. .... . .......
500 '
91.°c Clu*Irter2Pc'
Quatter1
g)tQuer ot Rand11.11
Assignment
Quarter 2 Quarter 3
Quarter Relative to Random Assignment
2 'Y
-214-
Quarter 4 Quarter 5
TABLE 8.6
SWIM
ALL AFDC -U: SHORT-TERM IMPACTS ON DISTRIBUTION OF EARNINGS,
EARNINGS/WELFARE MIX, AND MEASURED INCOME IN QUARTER FOUR
Employment and Welfare Outcomes Experimentals Controls Difference
Average Total Earnings,Quarter 4 (x)
None 61.4 66.9 -5.5**
$1 - $1500 16.8 14.6 +2.2
More Than $1500 21.9 18.6 +3.3
Total 100.0 100.0
Employment and Wel',ore Status,Quarter 4 (%)
Hod No Earnings, Received Same__AFDC Payments 44.2 51.0 -6.7***
Hod No Earnings, Received No__AFDC Payments 17.2 15.9 +1.3
Hod Some Earnings, Received Some__AFDC Toyments 20.5 16.4 +4.2**
Hod Some Earnings, Received NoAFDC Payments 18.0 p6.7 +1.3
Total 100.0 100.0
Average Measured Income,Quarter 4 (%)
None 17.2 15.9 +1.3
$1 $1500 18.5 16.J +2.2
More Than $1500 64.4 67.9 -3.5
Total 100.0 100.0
Average Measured Income (S)° 2040.14 2047.08 -6.94
Sample Size 687 654 1341
SOURCE AND NOTES: See Table 8.2.
280-215-
TABLE 8.7
SWIM
AFDC-U APPLICANTS AND RECIPIENTS: SHORT -TERM IMPACTSON EMPLOYMENT, EARNINGS, WELFARE RECEIPT, AND WELFARE PAYMENTS
Outcome and Fallow-Up Period
AFDC -U Applicants° AFDC-U Recipients°
Experimentals Controls Difference ! Experimentals Controls Difference
Ever Employed. Quarters 2-4 (%) 60.4 54.2 +6.2* 42.8 29.4 +13.4***
Average Number of Quarters withEmployment, Quarters 2-4 1.30 1.14 +0.16* 0.87 0.64 +0.23**
Ever Employed (%)Quarter of Random Assignment 47.4 45.4 +2.0 23.8 22.1 +1.7Quartor 2 41.3 36.9 ,4.5 28.2 18.4 +9.b***Quarter 3 44.5 38.9 +5.7* 27.9 21.5 +6.3*Quarter 4 44.5 38.5 +5.9* 30.9 24.3 +6.6*
Average Total Earnings.Quarters 2-4 ($) 3072.36 2723.59 +348.77 1306.36 1026.92 +279.44
Average Total Earnings (S)Quarrer of Random Assignment 821.75 8C .53 -45.78 271.81 210.77 +61.04Quarter 2 856.66 761.67 +95.00 352.73 285.05 +67.68Quarter 3 1104.46 945.26 +'19.20 433.17 334.02 +99.15Quarter 4 1111.24 1016.66 *r4.58 520.47 407.85 +112.62
Ever Received Any AFDC Payments.Quarters 2-5 ( %) 81.4 81.6 -0.1 93.1 93.7 -0.6
Average Number of Months ReceivingAFDC Payments. Quarters 2-5 6.35 6.55 -0.20 9.34 10.00 -0.66**
Ever Received Any AFDC Payments (%)Quarter of Random Assignment 78.1 75.7 +2.5 96.5 97.4 -0.9Quarter 2 77.6 77.2 +0.4 91.6 93.7 -2.1Quarter 3 57.8 58.8 -1.1 82.0 89.0 -7.0**Quarter 4 54.2 56.0 -1.8 79.9 84.4 -4.6Quarter 5 49.6 52.6 -3.1 74.9 77.7 -2.8
Average Total AFDC PaymentsReceived. Quarters 2-5 ($) 3704.47 3977.33 -272.86 6609.05 1235.80 -626.75***
Average AFDC Payments Received (3)Quarter of Random Assignment 784.81 815.17 -30.35 1966.85 1948.34 +18.50Quarter 2 1126.08 1150.75 -24.67 1851.68 1939.11 -37.43Quarter 3 888.24 953.97 -65.77 1644.82 1858.01 -213.18***Quarter 4 871.46 939.86 -48.40 1604.39 1775.10 -170.71*'Quarter 5 818.70 932.75 -114.06* 1508.16 1663.59 -155.43*
Semple Size 399 399 798 248 295 543
SOURCE AND NOTES: See Table 8.3. 281
AFDC-U applicants and recipients. Alt'ough there were earnings increases
for both groups, they were not statistically significant, partly due to
small sample sizes.
Among applicants, during the full follow -up period, 60.4 percent of
experimentals were employed at some point as compared to 54.2 percent c,f
the controls, a 6.2 percentage point increase. Statistically significant
employment gains occu.red during the last two quarters of follow-up as
well.
For recipients, during the entire follow-up period, the employment
gain was 13.4 percentage points more than the control group employment rate
oi 29.4 percent. These employment gains peaked in the second quarter at
9.8 percentage points, and then declined in quarters 3 and 4 to 6.3 and 6.6
percentage points, respectively.
While there were reductions in welfare receipt and payments for the
AFDC-U applicants, they were not statistically significant partly due to
small sample sizes. The only exception was a statistically significant
welfare saving of $114 in the final quarter of follow -up.
SWIM resulted in experimentals spending fewer months on welfare among
recipients. There were also reductions in the pro- -tion of experimentals
receiving welfare, but these were statistically significant only in the
third quarter. Both during the entire follow-up perioe and beginning in
the third quarter, there were also lower welfare payments for recipients.
E. Longer-Term Impacts
As with the AFDC analysis, two additional quarters of follow-up were
available for an early-enrolling sample of AFDC-U registrants. As
discussed in Chapter 2, there were differences between the early-enrolling
-217-
282
and later-enrolling samples which makes estimates for the early group
somewhat problematic as indicators of longer-term impacts for the whole
AFDC-U sample. The percentage of females was greater in the earlier sample
than the later one, and the average earnirgs of the earlier group were
low Jr than those of the later one. These samples did not differ, however,
in prior work history or welfare dependency. 7 Of particular concern are
the differences in impacts between the earlier and later enrolling samples
during the common short-term follow-up period. (See Appendix Table F.7.)
F...r example, employment rate gains and welfare savings appear to be larger
for the early-enrolling re,istrantr as compared to the later-enrolling
ones.
The impacts for the earlier sample, including two additional quarters
of follow-up, are shown in Table 8.8. Employment impacts were statisti-
cally significant beginning in the second quarter and continued through the
final quarter of follow-up. Earnings gains increased over time, but they
were statistically significant only in the final two quarters of follow-up.
This suggests that the employment and earnings impacts for the AFDC-U's may
continue beyo.td the sixth quarter, at least for the early-enrolling sample.
There were statistically significant reductions in the proporticn of
experimentals receiving welfare in the two additional quarters of follow-
up. Over the entire six-quarter follow-up period, tne early sample
received welfare for almost one month less than the control groi ".
Beginning with the third quarter, there were sustained welfare savings for
experimental s.
These preliminary findings provide evidence that impacts for AFDC-U
registrants will extend beyond the short-term period, although differences
2R 5-213-
7TABLE 8.8
SWIM
AFDC-U EARLIER COHORT: LONGER-TERM IMPACTSON EMPLOYMENT, EARNINGS, WELFARE RECEIPT, AND WELFARE PAYMENTS
Outcome and Follow-Up Period Experimentals Controls Difference
Ever Employed. Quarters 2-6 (%) 62.9 52.3 +1.).5***
Average Number of Quarters withEmployment, Quarters 2-6 2.01 1.59 +0.41***
Ever Employed (%)
Quarter of Rancom Assignment 38.5 36.7 +1.8Quarter 2 36.8 27.1 +9.7***Quarter 3 38.6 31.9 +6.8**Quarter 4 40.6 32.9 +7.6**Quarter 5 41.2 33.1 +8.1**Quarter 6 43.3 34.1 +9.2***
Average Total Earnings,Quarters 2-6 ($) 4275.58 3622.87 +652.71
Average Total Earnings ($)Quarter of Random Assignment 584.74 673.47 -88.73Quarter 2 654.88 585.96 +68.92Quarter 3 806.77 727.72 +79.04Quarter 4 841.76 753.33 +88.43Quarter 5 960.92 769.92 +191.00*Quarter 6 1011.26 785.94 +225.32*
Ever Received Any AFDC Payments,Quarters 2-7 (%) 86.1 86.5 -0.3
Average Number of Months ReceivingAFDC Payments, Quarters 2-7 10.36 11.31 -0.95**
Ever Received Any AFDC Payments (%)Quarter of Random Assignment 85.3 82.2 +3.2Quarter 2 82.3 83.2 -0.9Quarter 3 E5.3 71.5 -6.2*Quarter 4 63.4 68.2 -4.8Quarter 5 59.9 63.7 -3.8Quarter 6 52.0 59.7 -7.7**Quarter 7 50.4 57.5 -7.1**
(continued)
284-219-
TABLES 8.8 (continued)
Outcome and Follow-Up Period Experimentals Controls Difference
Average Total AFDC PaymentsReceived, Quarters 2-7 ($)
Average AFDC Payments Received ($)Quarter of Random AssignmentQuarter 2Quarter 3Quarter 4Quarter 5Quarter 6
Quarter 7
6621.93 7695.59 -1073.66***
1263.92 252.19 +11.741399.71 144,8.90 -69.191142.75 1328.52 -185.77***1124.01 1307.92 -183.91***1064.49 1246.38 -181.89**952.32 1192.56 -240.24***938.66 1151.32 -212.66***
Sample Size 375 348 723
SOURCE AND NOTES: See Table 8.4.
between the early and later samples necessitate particular caution in
making this inference.
Iv. Comparison of SWIM to Prior San Dieao Demonstration
In the early 1980's, San Diego operate a program that involved a
Short-term participation requirement of job search assistance and work
experience, referred to as EPP/EWBP. MDRC evaluated the effectiveness of
this program on AFDC and AFDC-U applicants and found that for the AFDC
applicants, there were employment and earnings gains as well as welfare
savings. For the AFDC-U's, this program reduced welfare benefits but did
not increase employment and earnings signficantly.8 An important question
is the extent to which a saturation program such as SWIM results in larger
cts than a more time-limited program such as EPP/EWEP. In making any
direct comparison of the SWIM findings reported in this chapter to those
presented in the final report of the earlier EPP/EWEP demonstration,
several differences should be noted:
First, the SWIM program served both applicant:. A recipients while
the EPP/EWEP program served only applicants. Second, operated in two
of San Diego's seven local offices while EPP/EWEP oper tel all seven
offices. 9Third, the research sample was followed froz ge point of
application at the Income Maintenance offices for the earlier demonstra-
tion, while for SWIM the research start date was the point at which
individuals showed up at the SWIM office; those who did not show up for
program orientation for such reasons as being denied aid or finding
employment are not included in the SWIM research sample, while they are in
the EPP/EWEP sample. Fourth, the economy during the period of the SWIM
-221- 28c3
program was better th:n during the earlier demonstration, which partly
accounts Zor a less imployable and more dependent group of
enrolling in SWIM as compared to EPP/EWEP.1°
The first three differences listed above can be control 'd for by
comparing the short-term impacts of (1) the applicant subgroup in SWIM to
(2) those who attended orientation in the two SWIM offices during EPP/EWEP.
Other differences between the two demonstrations, as well as the short-term
follow-up available for SWIM, do not allow these differences in the impacts
for the two demonstrations to be confidently attributed to variations in
the program model.
The two program modals resulted in fairly similar employment and
earnings gains and welfare savings for the AFDC applicants. For the AFDC-U
applicants, a similar comparison suggests SWIM may have resulted in larger
enpl:ymL...: and possibly larger earnings gains than EPP/EWEP, although
welfare savings are more similar.
Such comparisons, while based on experimental tests for each model
separately, do not use an experimental design to isolate the differential
impacts of the two models. It remains to be seen what effects longer-term
follow-up, or more sophisticated statistical techniques, will have on such
comparisons.
-222-
APPENDIX A
288
I
APPENDIX A
This appendix describes the County of San Diego's SWIM automated track-
ing system, how the data were used to produce the rates described in this
report, and how quality control procedures were carried out to verify the
accuracy of the data. Both county and MDRC staff invested substantial time
to ensure that all data items necessary for the research were collected and
data were of a high quality. Further, complicated programming was used to
produce the rates in the report.
It is Important to understand that the ease of calculating various
narticipation measures is affected by several factors: the type of partici-
pation measures required, the program model, and program resources.
In a complex program such as SWIM, it is inevitable that there will be
sane data quality problems since complex 'participant experiences' increase
the likelihood of error. However, the quality checks performed by MDRC
indicated that the data used in this report were of adequate quality to
produce reliable results. In fact, given the complexity of SWIM's program
model, it would have been difficult to further reduce the error rates
reported here.
This appendix summarizes the methods used to turn the information in
the SWIM autanated tracking system into a data set, and reports the primary
findings on the quality and completeness of this information. Sane general
considerations for management information systems designed to measure
participation are also given.
-224-
I. Description of the Partipation Data Set Used in the Report
The county's SWIM autanated tracking system provided most of the data
used in the analyses in Chapters 5, 6 and 7. Data were entered into the
system fran individual data input documents (see Figure A.1), completed by
CRU staff and supplemented by data fran program activity logs. A regis-
trant's 'experience in the program was defined by a) when they registered
for the program, b) when they started participating, c) when they finished
a component, and d) when they were deregistered. Consequently, these
documents were designed to track activities by identifying actual start and
end dates, as well as program status changes based on registration and
deregistration dates.
The top portion of the data input document tracked registrants' prog-
ram status (that is, availability or readiness for program participation,
registration, deregistration, deferrals, etc.). The remainder of the
document tracked activity referrals and outcomes (i.e., referrals,
no-shows, participation, component interruptions or completions). Cause
determinations for failure to meet program requirements and activity
provider codes were also tracked through the system. Every tracking form
was completed with a date documenting when the status change or activity
took effect.
The autanated tracking system maintained SWIM- related activities for
all the SWIM-eligibles; it was actually a record of .ates with associated
status or activity codes. Each time a staff verified or discovered that a
registrant's program or activity status changed, a new tracking document
would be completed and submitted to data entry. Note that staff were
required to update a registrant's tracking record for changes in status or
-225- 290
FIGURE A.1
SW IM
SW IM CLIENT TRACKING DOCUMENT
DEPARTMENT Of SOCIAL SERVICES
SWIM Client Tricking Document
NAME lard
Social Security No. I f I
rim en . 1.
PROGRAM STATUS: Detest /111El - REGISTERED/AVAILABLE (so reams required)
2 - DEFERRED R 2.10
0 3 - NON-PARTICIPATION lemons
4 warm PROGRAM CLOSURE reason:L. js DERECISTRATION rewrtL,
ACITITY STATUS CHANGES
(1)1ectivity code and dtle (gee list below)
I - REFERRED
2 - NO-SHOW FROM REFERRAL
3 - NON-PARTICIPATION
4 - ACTIVE TO COMPONT-NT
Remora (a INverem St.rus Change!
W - WIN criteria / exemptR - AFDC appUcatioo withdravrnD - AFDC &triedC - AFDC case closed - otterE - Employed 30 hrs/wk or moreT - Employed less than 30 lug /sek
F Reg. changed to rsaa-Fed.
provider code a-nd title (gee list below)
P- Excluded ParentU- Undoc. WorkerA - Appeal pending
X - DC!Vert/ tf 00 pendingS - Sanction spptledM - Reertratst movedN - No available activityB Ardgred to gods! wczker
to remove barriers
AG good canoe AN0 oo good cruse
AG good cause AN good came
5 - COMPONENT INTERRUPTION t /1_1_1 /1_1_,..t
6 - COMPONENT COMPLETED
(2) LIJ 1_4_1acdvity code and dtle (sec Urt below)
1 REFERRED
2 - NO-SHOW FROM REFERRAL
3 NON-PARTICIPATION
04 - ACTIVE TO coupon-NT
AG 0 good cause AN0 ro good cause
provider code gad title (tic I.12t below)
0 - COMPONENT INTERRUPTION .... 1 I t /1_1_3 it_j__1
6 - comFoNurr coNetrrED
Actavity codex Titles:
00 Coaroetlog0! - Employmem Search02 Job Search Woriohop03 Job Club04 MEP ()dentition
OS Work ExpedeoceOS - Vocedasal Testing07 - Remedied=011 - Clagertxrto Edocados09 51dIla Tralang
21 OJT22 Gnat Negritos
40. mplorosest referral41 employment 30 haheig
or lees
SOtsumWmSemployment more than
30 hrs/wk pendingderegistration
32-self-initiated edu-cation or training
AC good case AND us good came
AG good curse AN oo toad cause
AGO good ewe AND no good cause
Provider Codes G Titles:
CR . CADCW EWEPCP EPP
CS Sod al St tVi'MCD - JDUCC - actmr cocarty
SE MDSR ROI.
SS - other State of Celli.
F G Federal Cowl (!)FD - Federal Cov't (2)
JC - ITPA Cosonomity CollegeJT - ocher JTPA (1)
other JTPA (2)
PS Myatt SectorUN - Ulster's
LC coos aserdre/cablIc colleges GtrrdardonLP private fordo:dote
ET - ochry (1)KT - other (2)
TICKLER: 0 remove, or Diet to dale: s ellegt !
6211:0 OM pitted by:18-14 DSS (6/85)
daoe(6/86)
226-291
activity. However, some staff updated registants' records every 30 to 45
days, when they verified employment, training or education.
Two basic data sources were the minimum requirements for calculating
monthly participation rates: an indication of those individuals who were
registered as of a given point in time, and an indication of those
registrants who were also participating (that is, start and end dates of
all key components and statuses). These two sources required constant
updati?g.
In order to calculate the various monthly participation measures
discussed in this report, these start and end dates had to be converted
into a somewhat different format. First, several sets of monthly calendars
were established for each type of SWIM activity and program status. Then,
variables were set up that indicated each registrant's status in terms of
program eligibility and participation, as of each month in the
demonstration.
For example, consider a sample member who registered on September 15,
1986, participated in a job search workshop from October 1, 1986 through
October 15, 1986, found a job and was consequently deregistered on October
30, 1986. This sample member's 'program status' calendar would be coded to
reflect that she/he was registered in September and October, 1986, and not
during any other months of the demonstration. The months prior to and
following this period would be ccded to reflect that this person was not
registered. Similarly, the job search workshop and employment calendars
would be coded to reflect participation in these activities in October
1986. All the remaining calendars (e.g. EWEP, self-initiated and
program-referred training and education, etc.) would indicate no activity
-227- 292
for those components.
These calendars were used to calculate monthly participation rates, by
relying on the program status calendars to construct monthly denominators
of people registered and by relying on the activity calendars to construct
monthly numerators of people active in various components. Note that
special codes in the program status calendars or in calculations using
these calendars were required to accommodate various definitions of
SWIM-eligibles (such as 'registered as of the end of the month,' or
'registered throughout the mon'41').
In order to do the participation analyses, about 800 data items were
created from the raw data coded in the SWIM automated tracking system.
About 1,200 additional items were created during the course of the analysis
from this base (note this does not include outcome or demographic
variables). The large size of the data set required the use of a mainframe
computer for data manipulations and statistical calculations.
A great deal of county and MDRC staff time was spent monitoring the
original tracking data. Assembling and verifying the final data set was
time-intensive for MDRC staff as well. Calculations of consecutive monthly
participation rat_. would either not have been possible without the
County's automated tracking system or would have required considerably more
staff to do case file reviews.
Three factors will influence localities' ability-to calculate partici-
pation rates: resources available to create data systms or assign program
staff to this effort, the type of participation statistics desired, and the
complexity of the program model. If the program model is complicated, it
will be quite time consuming for program staff to collect data on all
ar)-228-
,
program components for all registrants: Staff would have to be familiar
with the varioui, codes for all components in the model; and programming to
calculate participation rates would require additional time and skill.
Longitudinal measures are easier to calculate and maintain in an autanated
system because no end dates are needed (relieving sane burden on staff);
monthly participation measures require end dates and more sophisticated
programming.
Periodic manual reviews of local office case files could be used to
calculate monthly participation rates. For such an effort to succeed, how-
ever, there must be an accurate and systematic method of identifying
registrants who were program-eligible in a month or quarter. Such a method
would have to involve complicated decision rules which would have to be
consistently 1,:atatained during very time consuming record reviews.
Examples of decision rules that would have to be set up include: How are
participants in more than one component coded? Are registrants with an
activity start date three weeks prior to the review but no end date or
ongoing verification code still considered participants? These decision
rules are more difficult to implement manually than by using computer
routines.
Even with an automated system, it was difficult for the participation
analysis reported here to distinguish between those who participated two
days out of a month and those who participated 15 days out of a month.
Capturing this participation on a daily basis would be very difficult to do
in the case file reviews as well as with an automated system.
The three factors that affect localities' ability to calculate parti-
cipation rates also affect the likelihood of introducing error into the
-229-
rates. The following section discusses the error rates found in the SKIM
autanated tracking system and steps that were taken to improve the quality
of these data for the research. That discussion also shows how program
models which require participation in a series of components require error
rates lower than the rates Shown in simpler program models in order to
maximize overall reliability in monthly or ongoing participation measures.
II. Quality of the Tracking Data Used in the Analysis
Many estimates presented in this report were sensitive to the quality
of the data obtained through the SWIM autanated tracking system. Conse-
quently, extensive checks were conducted to assess the accuracy and
completeness of the system's data and alternative data sources were used
where necessary.
Sane data quality problems are inevitable in any management informa-
tion system. The quality of the data collected for this study was similar
to (if not better than) that used in other MDRC studies of state welfare
initiatives. Thus, these are high quality program statistics. However,
the checks show that certain statistics in the report are more accurate
than others because the types of data used in the calculations had varying
degrees of accuracy. Consequently, there are' sane important caveats for
interpreting tables in the report. It is also important to note that the
quality checks reported here probably overstate the degree of unreliability
in le data used in this report. The quality check was conducted during
August, 1986; since then, county staff spent considerable time increasing
data acuracy.
MDRC's review of the SWIM autanated tracking system was complicated
-23
and comprehensive due to the many activities (and varying sources of data)
included in the SWIM evaluation. As stated above, it was critical for this
report to have accurate program status as well as activity start and end
dates. Depending on the component being checked, comparisons were made
between the data in the tracking system and the information found in the
registrant's case file or the individual component activity logs. Since
staff did not always enter a separate activity end date into the system for
active clients who deregistered, the quality of the deregistration data was
particularly important because this date was used as a proxy for activity
end dates in the analysis.
It is important to note that the SWIM tracking system input forms in
the case files were nct considered to be sufficient for verifying the dates
in the system. All information was confirmed on the basis of quality
checks that used supporting documentation independent of the system (e.g.,
case notes, transportation vouchers, documents fran service providers,
etc.). Several types of quality control checks were done. One compared
system data to case file or log data. Another checked system data against
other independent data sets, e.g., Unemployment Insurance, AFDC, etc.
Two skJecific data quality problems were investigated. One problem was
the extent to which dates in the automated system were incorrect. The
other was the extent to which dates that should have been in the system
were missing fran it. In calculating participation rates, both problems
could overstate or understate participation. Four estimates of quality
(where applicable) were made for each of the different components and
program statuses:
1) Of the start dates in the system, the percentage that were
-231-
286
accurate, according to the case files.
2) Of the end dates in the system, the percentage that wereaccurate.
3) Of those start dates that should have been in the system(those that were in the system plus those that were identifiedin the case file), the percentage that were missing.
4) Of those end dates that should have been in the system, thepercentage that were missing.
Data on most of the different types of components in the SWIM model
were checked: job search workshops, EWEP, job clubs, program-referred and
self-initiated education or training, and employment. In addition, data
concerning program statuses (such as registrations and deregistrations)
were checked. Table A.1 shows the error rates for various components as
well as program statuses (the latter are mostly entry and exit dates for
SWIM).
At least 90 percent of all program status data in the system were
confirmed as accurate. In addition, very few program status changes were
missing from the system: 1 percent of open entries (registrations,
re-registrations, etc.) and 5 percent of closure entries (deregistrations,
including sanctions, and closures due to registrants moving out of the SWIM
area) were missing from the system.
Since participation data for the job search workshops and job clubs
were keyed into the system directly from attendance logs, checks were not
conducted concerning the extent to which data in the system were accurate.
However, checks were done for missing data as shown in Table A.1.
There was sane underreporting of start dates for the job search work-
shops: 16 percent were missing in the system. Underreporting of end dates
was not as prevalent: only 11 percent of the job search workshops end dates
-23 2-2( .?
TABLE A.;
SWIM
SUMMARY STATISTICS FROM QUALITY CHECK OF DATAFOUND IN THE SWIM AUTOMATED TRACKING SYSTEM
Status or Activity Reviewed
Percei\t of Entries in
Systerr Confirmed asAccurcte
Program StatusaOpen Entries
bClosure Entries
Job Search WorkshopStort Dates
End Dates
EWEP
Start Dates in System Withino Month of Initial
ParticipationStart Dates In System Any
Time Before or AfterInitial Participation
End Dates for IndividualsWith Stort Dates
Job ClubParticipation in o
Particular Session
Education an TrainingStort DatesEnd Dates for IndividualsWith Start Dates
Employm.atStart DatesEnd Dates
Percent of KnownEntries Missing fromSystem
92 (93/101) 1 (1 /1D2)
90 (38/42) 5 (2/44)
N/A 16 (15/94)N/A 11 (10/94)
N/A 26 (28/106)
N/A 22 (23/106)
N/A 4 (2/48)
N/A 11 (4/36)
76 (13/17) N/A
92 (12/13) 0 (0/13)
1DD (23/23) 15 (4/27)92 (11/12) 8 (1/13)
NOTES: Numbers in parentheses refer to the numbers used in the percentcalculatiores.
N/A indicates not applicable. Data was keyed directly into thetracking system from attendance logs, so 'false. participation was not expected.Therefore, quality checks were not completed.
aOpen entries include registrations, re-registrations, deferrals and
non participation statuses.
bClosure entries include moving out of the SWIM area, deregistrations
and sanctions.
233293
were not entered into the system. To correct for these missing end dates,
proxy end dates -- defined as 30 days from the workshop start date -- were
created. Participation in job clubs was not entered into the system for 11
percent of the sessions attended.
The error rates found for EWEP information in the automated system
were considered too high to produce accurate calculations. Consequently,
EWEP attendance logs were used as the research data source for EWEP partici-
pation, instead of the automated tracking system data.
Only 76 percent of all the education and training start dates that
were entered into the system were confirmed by information in the case
files. This may suggest that up to one-quarter of participation in educa-
tion and training reported in the system may have been 'false,' possibly
causing over estimates of participation in this report. This is probably
an overestimate of the error rate, however, since in some cases staff only
used the SWIM automated tracking system to keep track of these activities,
instead of adding supplemental documentation to the case files. This made
comparisons of education and training in the SWIM tracking system to
evidence of education and training in the case files difficult.
The length of education and training participation appeared to be
accurate in the system. That is, it was not overreported due to missing or
inaccurate end dates. Ninety-two percent of those with start dates had
accurate end dates in the system.
All employment start dates that were entered in the system were
verified as accurate through case file reviews. Additionally, all the
employment codes entered in the system had corresponding end dates. These
reviews indicated, however, that overall employment was underreported by 15
-234-
percent in the automated system when compared to the information in the
case file. Although employment may be underreported (by 15 percent) in the
tracking system, reliable end dates indicate that the length of employment
was not overreported.
Fluctuation in the number of, hours registrants worked may have
accounted for sane of the employment coding and entry errors. The quality
control effort indicated, however, that this was not a large problem. For
example, staff might have part-time employment in the tracking system for
one month, then full-time employment the next month if the hours increased
to 30 or more. However, staff may easily have forgotten to enter an end
date for part-time employment as the employment itself was not new.
That the SWIM automated tracking system was of adequate qiiality to
produce reliable and accurate statistics for analysis (once EWEP attendance
log data was merged with it) is confirmed by comparing the error rates
found in similar quality checks for two other MDRC work/welfare evalua-
tions. In the California EPP/EWEP demonstration, case files confirmed 85
percent of the transactions reported in the EPP Information System (EPPIS).
However, EPPIS was missing 8 percent of the job search workshop participa-
tion, 12 percent of the employment and 25 percent of the deregistration
activity indicated in the case files. In the West Virginia demonstration,
100 percent of the WIS tracking transactions were confirmed in the case
files for most types of transactions: work experience, individual job
search, employment and deregistrations. In addition, the system was found
to be missing only 3 percent of the work experience transactions, 7 percent
of the employment transactions, and 3 percent of the deregistration
transactions.
'00-235-
The SWIM model had more components for which tracking data were re-
quired than was the case in the other two demonstrations. The need to take
into account several different program components could have compounded the
individual component error rates for long term participants and introduced
an overall error rate greater than the error rate of a single component.
This is an issue that should be kept in mind when gathering data on
programs with complicated models.
The following example, not typical of most participants, shows has the
error rates of separate components can lead to a higher chance of a parti-
cipant's experience being inaccurate. Some individuals might have gone
through a job club, found part-time employment, and then participated in a
job search workshop. Since there was an 11 percent chance of missing the
participation in the job club and an 8 percent chance of missing an end
date for employment, by the time a person got to the job search workshop
after employment, there was an 18 percent chance that this person's 'SWIM
experience' was incorrectly documented: either incorrectly in employment or
incorrectly still in the job club while employed [(0.08 + 0.11)-(0.08x0.11)
or (0.08x0.89) + 0.11]. Note that this particular pattern is only relevant
for those few individuals who went through the three most error-prone
components before deregistering. In any given month, however, the error
rate will be considerably lower than 18 percent because other components
have lower error rates and registrants are at different points in their
program tenure.
This example shows that the likelihood of peoples' tracking history
being wrong was higher the more components they went through. The error
rates in management information systems must be particularly low in order
3 01-236-
to accurately measure the many different aspects of a complicated program
model (as is the case with SWIM). A less complicated program model would
not require individual component error rates to be so low because the
errors would not be compounded.
As a second type of check for accuracy, data from the SWIM automated
tracking system was compared to the Unemployment Insurance earnings and
AFDC payments records data.
The results of the Unemployment Insurance match indicated the extent
to which the employment data in the tracking system were confirmed. (The
extent to which employment data were missing from the tracking system could
not be ascertained using the Unemployment Insurance system.)
For example, 28 percent of those AFDC's who were employed within 12
months according to the tracking system were not employed according to the
Unemployment Insurance system. A similar rate was found for AFDC-Ws.
This is not surprising since much of this discrepancy may be because there
are sane jobs not covered by the UI system.
Registration data from the SWIM automated tracking system were com-
pared against AFDC payment records as another test of data quality. This
check showed that there were people who were registered with SWIM according
to their tracking records, but who did not receive an AFDC payment in a
given month. In August, 1986, for instance, 9 percent of those who were
registered at least three weeks during the month (AFDC's and AFDC-U's
combined) had no AFDC payments during that month. Since applicants denied
AFDC, zero grants due to temporary high earnings or simple communication
delays between the IM and SWIM offices can be expected to account for a
-237-
31)2
discrepancy of this size, it confirms the high quality of the registration
data.
III. Implications for Management Information Systems
Chapter 5 focused on longitudinal activity measures: the extent to
which individuals ever participated in the program within 12 months. These
measures relied only on activity start dates. The ongoing and monthly
participation measures discussed in Chapters 6 and 7, however, required
both activity start and end dates in order to accurately calculate
participation.
If performance criteria are mandated, they may have less chance of
error if they do not rely on both start and end dates, or on the separate
calculation of different components. Longitudinal activity measures (such
as Table 5.3) are more reliable than the monthly participation measures
because they do not rely on end dates (such as those shown in Figure 7.2).
A program model that includes components that are difficult to monitor
(such as employment, training or education) also introduces more room for
error than a model that tracks only program-operated components.
As discussed in Chapters 5, 6 and 7, participation rates vary depend-
ing on the method of calculation (longitudinal versus monthly participation
measures). These different methods of calculating participation as well as
the program model may introduce different error rates and different burdens
on program staff.
In summary, the SWIM demonstration effectively implemented a manage-
ment information system for producing reliable participation measures in a
complex program model, even though it was developed primarily for research
-238- 300
purposes. Data sets on program activities will rarely be more reliable.
Although SWIM's experience indicates that an adequate system to measure
program activity rates can be developed, it also makes it clear that such
an effort requires considerable planning and major resources.
304-23 9-
TABLE B.1
SWIM
SELECTED CHARACTERISTICS OF REGISTRANTS
AT THE TIME OF INITIAL REGISTRATION, BY ASSISTANCE CATEGORY AND RESEARCH GROUP
Charocteristic
AFDC AFDC-U
Experimental Control Experimentll Control
AFDC Status (%)
Applicant 40.2 38.3 58.1 61.5
Renewed Recipient 33.5 32.3 23.7 22.5
Redetermined Recipient 26.3 29.4* 18.2 16.0
Averoge Age (Yeors) 34.1 34.3 32.7 33.0
Sex (%)
Male 8.8 8.6 92.0 90.5
Femole 91.2 91.4 8.0 9.5
Ethnicity (%)
White, Non-Hispanic 28.2 26.3 25.0 24.5
Block, Non-Hisponic 42.0 42.3 21.6 18.6
Hispanic 25.7 25.6 40.8 43.5
American Indian/Aloskan Native 0.4 0.7 0.1 0.7
Asion and Pacific Islander 2.9 4.6** 11.1 11.1
Other 0.7 0.4 1.4 1.6
Degree Received (%)
High School Diplomo 48.1 48.0 39.0 36.7
GED 8.1 7.5 6.8 9.2
None 43.8 44.5 54.2 54.0
Averoge Highest Grade Completed 10.9 10.9 10.2 10.0
Current Activities (%)ex
Employed 20 Hours or Less Per Week 7.4 6.6 6.7 6.0
Employed 21-30 Hours Per Week 5.4 5.6 4.3 1.9**
Education or Training 14.3 14.8 9.7 9.5
Priar AFDC Dependency (%)
Never on AFDC 11.9 10.9 34.5 33.7
1-11 Months 6.4 7.5 15.1 15.8
12-23 Months 7.0 6.1 10.9 10.2
24-35 Months 8.4 7.8 11.4 10.1
36-47 Months 8.4 8.3 8.2 7.5
48-59 Months 6.8 7.0 6.n 7.6
60 Months ar More 51.1 51.8 13.9 15.1
(continued)
-242-306
TABLE B. (continued)
Characteristic
AFDC AFDC-U
Experimental Control Experimental Control
Average Number of Months on AFOC During
24 Months Prior to Initial Registration 15.5 15.5 9.6 9.4
Held a Job at Any Time During Quagter
Prior to Initial Registration (s) 26.6 26.9 37.5 38.8
Held a Job at Any Time During
Four Quarters Prior to
Initial Registration (%) 38.9 39.9 56.9 56.1
Average Earnings During Quatter Prior
to Initial Registration ($) 415.69 428.00 920.78 818.18
Average Earnings During Four Quarters
Prior to Initial Registration ($) 1650.31 1686.85 3782.15 3217.19*
Received Unemployment Compensation
During Three Months Prior to Initial
Registration (s) 4.0 4.4 9.9 8.4
Average Amount of Unemployment
Compensation During Three Months
Prior to Initial Registration ($) 30.10 33.99 67.83 69.80
Sample Sizec 1608 1619 704 683
1986.
SOURCE: See Table 2.3.
NOTES: The somple for this table includes individuals who registered between July 1985 and June
Distributions may not add to 100.00 percent due to rounding.
A chi-square test or t-test was applied to differences between experimental and controi
groups within assistance categories. Statistical significance levels are indicated as: * = 10 percent; sue
5 percent; *** = 1 percent.
aAFDC-U cases can be redetermined as WIN-mandatory when an AFDC case becomes an AFDC-U case or
when a previously exempt AFDC-U case (e.g., medically exempt) loses its exemption status.
bThese data are calculated from the State of California Unemployment Insurance earnings
records and include zero values for sample members not employed and for those not receiving Unemployment
Compensation.
c
For selected characteristics, sample sizes may vary up to 5 sample points due to missingdata. 62 of these registrants were excluded from the impact analysis because they did not have social
security numbers.
-243- 30 7
TABLE B.2
SWIM
SELECTED CHARACTERISTICS OF REGISTRANTS AT Tilt TihIE OF INITIAL REGISTRATION,
BY ASSISTANCE CATEGORY AND PERIOD OF INITIAL REGISTRATION
Characteristic
AFDC AFDC -U
Earlier Cohort Later Cohort Earlier Cohort Later Cohort
AFDC Status (%)
Applicant 42.8 35.0rn 60.0 59.5
Renewed Recipient 29.2 37.3 * ** 20.2 26.6***
Redetermined Recipient 28.0 27.7 19.8 13.91011
Average Age (Years) 33.8 34.5** 32.5 33.1
Sex (%)
Mole 8.9 8.6 89.2 93.7***
Female 91.1 91.4 10.8 6.3 * **
Ethnicity (%)
White, Non-Hispanic 27.3 27.1 25.0 24.4
Black, Non-Hispanic 42.1 42.3 20.6 19.5
Hispanic 26.0 25.2 42.6 41.6
American Indian/Alaskan Native 0.6 0.5 C.4 0.5
Asian and Pacific Islander 3.2 4.4 9.6 12.9*
Other 0.7 0.4 1.9 1.1
Degree Received (%)
High School Diploma 48.7 47.3 37.0 38.9
GEO 8.4 7.1 7.7 8.3
None 43.0 45.6 55.3 52.8
Average Highest Grade Completed 10.9 10.9 10.1 10.1
Current Activities (%)
Employed 20 Hours or Less Per Week 7.6 6.2 6.5 6.1
Employed 21-30 Hours Per Week 5.7 5.2 3.5 2.7
Education or Training 12.8 16.7 * ** 9.6 9.6
Prior AFDC Dependency (%)
Never on AFDC 12.0 10.7 35.2 32.8
1-11 Months 7.5 6.2 16.2 14.5
12-23 Months 7.0 6.7 10.5 10.7
24-35 Months 8.3 7.8 10.4 11.2
36-47 Months 8.9 7.7 8.4 7.2
48-59 Months 7.2 6.6 6.3 7.4
60 Months or More 49.1 54.3*** 13.0 16.2
(continued)
308-244-
TABLE 8.2 (continued)
Chorocterl sti c
AFDC AFDC-U
Earlier Cohort Later Cohort Earlier Cohort Later Cohort
Average Nunber of Months on AFDC During24 Months Prior to Initial Registration 15.2 16.0** 9.4 9.7
Held a Job at Any Time During QuotterPrior to Initial Registration (s) 27.9 25.4 36.4 40.1
Held a Job at Any Time During FourQuarters Prior tg InitialRegistration (X) 39.4 39.4 54.2 59.2
Average Earnings During Quotter Priorto Initial Registration ($) 458.61 377 .37 ** 830.29 917 .92
Average Earnings During Four QuartersPrior to Initial Registration (s) 1583.10 1771.87 3106.59 3977.22**
Received Unemployment Canpensati onDuring Three Months Prior to InitialRegistration (X) 4.4 3.9 8.3 10.2
Average Amount of UnemploymentCompensation During Three MonthsPrior to Initial Registration (0 34.01 29.67 62.83 75.76
ample Sizec 1769 1458 752 635
SOURCE: See Table 2.3.
NOTES: The earlier cohort registered between July 1985 and December 1985 and the later cohortregistered between January 1986 and June 1986.
Distributions may not add to 100.0 percent due to rounding.
A chl-square test or t-test was oppl led to differences between cohorts within assistancecategories. Statistical significance level s are Indicated as: = 10 percent; = 5 percent ; = 1 percent.
aAFDC-U cases can be redetermined as WIN-mandatory when an AFDC case becomes an AFDC-U case or
when a previously exempt AFDC-U case (e.g.. medically exempt) loses ISM exemption status.
bThese data are cal culated from the State of California Unemployment insurance earnings records
and Include zero values for sample members not employed and for those not receiving Unemployment Compensation.
cFor selected characteristics, sample sizes may vary up to 5 sample points due to missing data.
62 of these registrants were excluded from the impact analysis because they did not have social securitynumbers.
245 303..
TABLE 8.3
SELECTED :HARACTERISTICS OF SWIM-ELIGIBLES AT THE TIME OF INiTIAL REGISTRATION,
BY ASSISTANCE CATEGORY AND YEAR OF INITIAL REGISTRATION
Characteristic
AFDC AFDC -U
July 1985-
June 19S6
July 1986 -
June 1987
July 1985-
June 1986
July 1986 -
June 1987
AFDC Status (%)
Applicant 40.2 58.2*** 58.1 80.6***
Renewed Recipient 33.5 7.1 * ** 23.7 7.1 * **
Redetermined Recipient 26.3 34.7*** 18.2 12.2**
Average Age (Years) 34 1 32.0*** 32.7 I 31.4**
Sex (%)
Male 8.8 11.0 92.0 92.9
Female 91.2 89.0 8.0 7.1
Ethnicity (%)
White, Non-Hispanic 28.2 30.2 25.0 28.6
Black, Non-Hispanic 42.0 40.6 21.6 18.4
Hispanic 25.7 24.5 40.8 41.3
American Indian/Alaskan Native 0.4 0.7 0.1 0.3
Asian and Pacific Islander 2.9 2.9 11.1 10.5
Other 0.7 1.1 1.4 1.0
Oegree Received (%)
High School Diploma 48.1 50.5 39.0 41.6
GED 8.1 8.2 6.8 10.7**
None 43.8 41.3 54.2 47.7**
Average Highest Grade Completed 10.9 11.2** I 10.2 I 10.4
Current Activities (%)
Employed 20 Hours or Less Per Week 7.4 3.5*** 6.7 6.9
Employed 21-30 Hours Per Week 5.4 4.0 4.3 1.5*Education or Training 14.3 15.5 9.7 7.7
Prior AFDC Dependency (%)
Never on AFDC 11.9 .i**: 34.5 54.3***
1-11 Months 6.4 11.2*** 15.1 20.2**12-23 Months 7.0 7.7 10.9 7.1*
24-35 Months 8.4 6.9 11.4 6.1***
36-47 Months 8.4 6.2* 8.2 4.1**48-59 Months 6.8 6.6 6.0 1.3 * **
60 Months or More 51.1 40.2*** 13.9 6.9***
(continued)
3.10-246-
-__
TABLE 8.3 (continued)
Choracteristic
AFDC AFDC -U
July 1985 -
June 1986
July 1986 -
June 1987
July 1985 -
June 1986
July 1986 -
June 1987
Average Number of Months on AFDC During
24 Months Prior to Initiol Registrotion 15.5 11.1*** 9.6 3.9**s
Held a Job at Any Time During Quarter
Prior to Initiol Registration (%) 26.6 20.2*** 37.5 34.4
Held a Job of Any Time During Four
Quorters Prior to Initio" 38.9 29.6*** 56.9 48.O***Registration (%)
b
Average Earnings During Quarter Prior
to Initiol Registration (s) 415.69 325.06 920.78 862.31
Average Earnings Ourlag Four Quorters
Prior to Initial Registration (s) 1650.31 1503.26 3782.15 4260.63
Received Unemployment Compensation
During Three Months Prior to Initial
Registration (%) 4.0 4.6 9.9 14.4**
Average Amount of Unemployment
Compensation During Three Months
Prior to Initiol Registrotion (5) 30.10 36.30 67.83 145.38***
Sample Sizec 1608 820 704 392
1987.
SOURCE: See Table 2.3.
NOTES: The sample for this table includes individuols who registered between July 1985 and June
Distributions may not add to 100.0 percent due to rounding.
A chi-square test or t-test wos applied to differences between the July 1985 through June
1986 and the July 1986 through June 1987 registrant groups within assistance categories. Stotistical
significance levels are indicated os: * = 10 percent: *S = 5 percent: *** = 1 percent.
°AFDC -U cases can be redetermined as WIN-mondatory when an AFDC case becomes an AFDC-U case
or when a previously exempt AFDC-U case (e.g., medical]] exempt) loses its exemption status.
bThese data are calculated from the Stote of California Unemployment Insurance earnings
records and include zero values for sample members not employed and for those not receiving Unemployment
Compensation.
doto.
c
For selected choracteristics, sample sizes may vary up to 5 sample points due to missing
311
APPENDIX C
312
TABLE C.1
SWIM
TWELVE- AND EIGHTEEN-MONTH ACTIVITY MEASURESOR SWIM-ELIGIBLES, BY ASSISTANCE CATEGORY
Activity Measures
12-Month Indicators 18-Month Indicators
AFDC AFDC-U AFDi: AFDC-U
Participated in Any Component, IncludingEmployment While Registered 78.0% 72.5% 79.2% 73.0%
Participated in Any Component, ExcludingEmployment While Registered 65.3 63.4 67.2 64.2
Participated in Job Search Activities 52.4 55.6 54.8 56.6
Participated in Work Experience 20.9 21.0 23.0 /2.3
Participated in Education or Training 23.9 15.3 27.2 17.7
Employed While Registereda 40.4 33.0 42.9 36.4
Moved Out of the SWIM Area 8.1 8.8 8.9 9.6
Deregistered 63.6 67.3 72.5 75.6Due to Sanction 9.8 7.3 10.6 8.1
Sample Size 874 385 374 385
SOURCE: MDRC calculations from the County of San Diego Department of SocialServices SWIM Automated Tracking System and EWEP attendance logs.
NOTES: The sample for this table consists of individuals woo registeredbetween July and December 1985.
Activity measures are calculated as a percentage of the totalnumber of persons in the indicated assistance category. Fallow-up periods-beginat the point of Initial registration.
Participation is defined as attending EWEP for at least one hour orany other activity for at least one day.
Tests of statistical significance were not examined.
aProgram employment information is based on employment that was
reported to program staff. Program employment data were not used to measureimpacts.
313-250-
APPENDIX D
3 1 4
TABLE D.1
SWIM
PERCENTAGE DISTRIBUTION OF SWIM-ELIGIBLES,
BY NUMBER OF MONTHS ACTIVE OUT OF MONTHS REGISTERED
DURING THE TWELVE MONTHS FOLLOWING INITIAL REGISTRATION
Number of
Months Active
0
1
2
3
4
5
6
7
8
9
10
11
12
Total
Number of Month.: Registered
1 2 3 4 5 6 7 I 8 9
84.3
15.7
100.0
63.2 48.3
16.6
20'.3
17.2
19.3
15.2,
100.0 100.0
32.3
14.9
15.4
y10:1-
18.2
100.0
21.2
12.5
13.3
14.3
20.7,
100.0
20.8
12.1
15.3
18.0
6.9
'13.1.
13:9:
100.0
Sample Size 93 213 214 149 167 119
10.0
5.0
9.8
7.4
7.2
11.4
15.7
8.5
104 I 110
'10.5
100.0
106
10 11 12 Total
9.0 6.6 9.5 24.4
4.1 1.1 3.6 8.8
13.3 5.7 3.0 9.7
8.5 3.8 2.9 7.5
6.7 4.9 3.3 5.7
9.0 7.4 6.2 6.9
8.7 12.8 6.1 5.3
15.0 9.0 6.4
9.7 5.5
E
14.4 4:8 4.8:
5.9 (10:9' 0:4 4.1
! 10.i . 12:9. 5.2
15.6 5.8
100.0 100.0 100.0 100.Q
90 85 860 2312
SOURCE: MDRC calculations from the County of San Diego Department of Social Services SWIM Automated
Tracking System and EWEP attendance togs.
NOTES: The sample for this table includes individuals who registered between July 1985 and June 1986.
The sample is weighted to adjust for the higher proportion of AFDC's, relative to the proportion
of AFDC -U's, in our sample.
months.
The shaded area represents individuals who participated in at least 70 percent of their registered
3 1 5
-252-
(continued)
least one day.
TABLE D.1 (continued)
Distributions may not add to 100.0 percent due to rounding.
Participation is defined as attending EWEP for at least one hour or any other activity for at
Registered during a month is defined as registered for at least one day during the month.
Tests of statistical significance were not examined.
-253-
3 i 6
TABLE D.2
SWIM
PERCENT OF NONCOMPLIANT SWIM-ELIG1BLES RECEIVING FOLLOW-UP CONTACT,BY METHOO OF FOLLOW-UP ANO ASSISTANCE CATEGORY
Follow-Up Method AFDC AFOC-U Total
Telephone Call 18.3% 21.5% 19.6%
Automatic Reschedule (byletter), No Oirect ContactWith Registrant 27.0 29.1 27.8
Reschedule (by letter),After Direct Contact withRegistrant 37.4 45.6 40.7
Letter Sent to Set UpMeeting With Registrant 40.0 40.5 40,2
Meeting with Registrant inOffice 20.0 27.9 23.2
Referral to WelfareEligibility Technician 16.5 12.7 15.0
Home Visit 1.7 3.8 2.6
Registrant Initioted Contact 50.4 36.7 44.9
Staff-Registrant ContactInitiator Unknown 13.0 19.0 15.5
Sample Size 115 79 194
SOURCE: MDRC calcui3tions from cosefile reviews of randomly chosenSW IA-Eligibles who registered between January 1 and March 31, 1986.
NOTES: Oistributians may not add to 100.0 percent because more than onemethod of follow-up may have been used for each individual.
Noncompliance is defined as falling to attend an assigned activityfor the required number of days or failing to meet other program requir .ents.Other requirements include verifying education or training attendance ..dbehaving in a cooperative manner while attending a program component.
Tests of statistical significance were not examined.
337254
TABLE D.3
SWIM
PERCENTAGE DISTRIBUTION OF AFDC SWIM-ELIGIBLES, BY WELFARE, PROGRAM,EMPLOYMENT, AND PARTICIPATION STATUS BY THE TWELFTH MONTH
FOLLOWING INITIAL REGISTRATION
Welfare, Program and EmploymentStatus Participants
Non-Participants Total
All AFDC
On WelfareDeregistered
Employed 5.4 2.7 8.1Not Employed 17.1 10.4 27.5
RegisteredEmployed 6.4 2.1 8.5Not Employed 16.5 2.6 19.1
Off Welfare (Deregistered)Employed 9.7 7.1 16.8Not Employed 10.5 9.4 19.9
Total AFDC Sample 65.6 34.3 100.0
AFOC Applicants
On WelfareDeregistered
Employed 5.5 3.0 8.5Not Employed 13.3 10.1 23.4
RegisteredEmployed 4.3 2.1 6.4Not Employed 8.7 1.8 10.5
Off Welfare (Deregistered)Employed 11.6 10.7 22.3Not Employed 13.9 14.9 28.8
Total AFDC Applicants 57.3 42.6 100.0
AFDC Recipients
On WelfareDeregistered
Employed 5.4 2.5 7.9Not Employed 19.7 10.5 30.2
RegisteredEmployed 7.9 2.1 10.0Not Employed 21.9 3.3 25.2
Off Welfare (Deregistered)Employed 8.4 4.6 13.0Not Employed 8.1 5.5 13.6
Total AFDC Recipients 71.4 28.5 100.0
Sample Sizes: All AFDC 892 468 1360Applicants 322 240 562Recipients 570 228 798
3 1
-255-
(Continued)
TABLE D.3 (continued)
SOURCE: MDRC calculations from the County of San Diego AFDC records, theState of California Unemployment Insurance earnings records, the County of San DiegoDepartment of Social Services SWIM Automated Tracking System, and EWEP attendancelogs.
NOTES: The sample for this table consists of SWIMEligibles who registeredbetween July 1985 and April 1986.
All percentages were calculated as a proportion of the total number
of SWIMEligibles, not as a proportion of the total number of Participants orNonParticipants.
Distributions may not sum tc totals due to rounding.
Participation is defined as having attended EWEP for at least one
hour, or any other activity for at least one day.
Unlike other tables, individuals who were off welfare by the twelfthmonth following initial registration were considered to be deregistered. In other
tables, deregistration is defined as being deregistered according to the County of
San Diego Department of Social Services SWIM Automated Tracking System.
Tests of statistical significance were not examined.
-256-
TABLE 0.4
SWIM
PERCENTAGE DISTRIBUTION OF AFDC -U SWIM-ELIGIBLES, BY WELFARE, PROGRAM,EMPLOYMENT, AND PARTICIPATION STATUS BY THE TWELFTH MONTH
FOLLOWING INITIAL REGISTRATION
Welfare, Progrom and EmploymentStatus Participants
Non-Participants Total
All AFDC -U
On WelfareDeregisteredEmployed 8.5 3.9 12.4Not Employed 16.2 7.5 23.7
RegisteredEmployed 3.8 0.5 4.3Not Employed 13.0 3.2 16.2
Off Welfore (Deregistered)Employed 12.3 9.2 21.5Not Employed 11.5 10.3 21.8
Total AFDC-U Sample 65.3 34.6 100.0
AFDC -U Applicants
On WelfareDeregisteredEmployed 7.0 4.1 11.1Not Employed 16.3 5.8 22.1
RegisteredEmployed 3.5 0.3 3.8Not Employed 7.6 1.5 9.1
Off Welfare (Deregistered)Employed 15.7 12.8 28.5Not Employed 12.8 12.8 25.6
Total AFDC -U Applicants 62.9 37.3 100.0
AFDC -U Recipients
On WelfareDeregisteredEmployed 10.8 3.7 14.5Not Employed 16.2 10.0 26.2
RegisteredEmployed 4.1 0.8 4.9Not Employed 20.7 5.8 26.5
Off Welfare (Deregistered)Employed 7.5 4.1 11.6Not Employed 9.5 6.6 16.1
Total AFDC -U Recipients 68.8 31.0 100.0
Sample Sizes: All AFDC -U 382 203 585Applicants 216 128 344Recipients 166 75 241
SOURCE and NOTES: See 'able 0.3.
320-257-
APPENDIX E
3r) ,4., _L
TABLE E.1
SWIM
PERCENT OF INDIVICUALS ELIGIBLE FOR SWIM IN JULY OR NOVEMBER 1986
WHO PARTICIPATED IN PROGRAM ACTIVITIES, BY
TYPE OF ACTIVITY AND ASSISTANCE CATEGORY
Activity
July 1986 November 1986
AFDC AFDC-U Total AFDC AFDC-U TOtal
Participated in Program-Arranged
Activities 27.4% 29.1% 28.0% 21.3% 21.4% 21.3%
Work Experience° 4.3 5.1 4.6 5.8 4.7 5.4
Job Search Activities 17.1 20.1 18.1 9.6 11.6 10.3
Job Search Workshop 8.0 9.3 8.4 4.2 4.1 4.2
Job Club 7.2 8.6 7.6 4.0 5.0 4.3
ISESA 1.9 2.2 2.0 1.5 2.5 1.8
Program-Arranged Education or
Training 6.0 4.0 5.3 5.9 5.1 5.6
Participated in Self-nitioted
Education or Training 7.4 7.5 7.4 12.2 10.2 11.5
Employed While Registeredc 20.4 18.1 19.6 20.7 15.6 19.1
Somple Size 2114 1059 3173 2326 1142 3468
SOURCE: ADRC calculations from the County of San Diego Department of Social Services SNIM
Automated Tracking System ond ENEP attendance logs.
NOTES: The sample for this table includes Individuals who were registered for SWIM In July
or November 1986.
The sample Is weighted to reflect the octuol number of SWIM - Eligibles.
Participation Is defined os attending ENEP for at leost one hour or any other
activity for of least one doy.
Participants who were active In more thon one activity were counted In the cotegory
with the highest priority. The priority order is: ENEP or OJT; job search workshop; job club;
ISESA; program-orronged education or training; self-initioted education or training ond union or
other job seorch; employment while registered.
The total monthly participation rote may be obtained by adding the participation
rates for program-orronged activities (which include the sum of work experience, job search
activities and education or training); self-initiated education or training and union or other job
seorch; ond employment.
(continued)
-260-
TABLE E.1 (continued)
Tests of stotisticol signif iconce were not examined.
oWork experience includes EMI EP and On-the-Job Troining.
b This cotegory also Includes o few individual s who participated in union and othertypes of self-Initioted job search octivi ties.
c Program employment informotion is based on employment thot wos reported to programstoft. Program employment dato were not used to measure Impacts.
323
TABLE F.1
SUM
AFOC: ESTIMATED REGRESSION COEFFICIENTS FOR
EMPLOYMENT ANO WELFARE MEASURES IN QUARTER FOUR
Variable
Variable
Mean
Ever
Employed (%) Earnings (8)
Received
AFOC ( %)
AFOC
Payments ($)
Experimental Group Member .500 +0.078*** +118.12*** -0.055*** -128.27***
(0.015) (40.16) (0.014) (24.43)
Office
Son Olego West .501
Service Center .499 +0.014 +25.20 +0.027* +43.11*
(0.015) (41.55) (0.015) (25.28)
Age Greater Thon or .459 -0.013 +54.54 -0.020 -94.02***
Equal to 35 (0.016) (44.03) (0.016) (26.79)
Female .913 +0.068** +150.60** +0.060** +41.28
(0.028) (76.61) (0.027) (46.62)
High School Diploma or GEO .561 +0.064*** +188.77*** -0.040** -97.13***
(0.017) (44.76) (0.016) (27.23)
Morirol Status
Married .334 - - -
Never Married .301 +0.032 +87.92 +0.016 -65.57*
(0.021) (55.66) (0.020) (33.86)
Divorced/Widowed .365 +0.047*** +159.38*** +0.005 -58.64*
(0.018) (49.28) (0.017) (29.98)
Family Status
Any Children Less Thon 6 .100 -0.009 +10.51 +0.088*** +271.71***
(0.029) (78.04) (0.023) (41.48)
Any Children 6 to 18 .904 -0.019 +105.79 +0.014 +73.98
(0.032) (87.64) (0.031) (53.33)
Race/Ethnicity
Whim, Non - Hispanic .322 - - - -
Block, Non-Hisponic .424 -0.034* -65.91 +0.047*** +115.67***
(0.018) (49.59) (0.018) (30.17)
Hispanic .254 -0.010
(0.021)-32.15
(51.20)
+0.031
(0.020)
+86.67**
(34.80)
(continued)
-264-
TABLE F.1 (continued)
VariableVariable
jean
Ever
Employed (%) Earnings ($)
Receiveo
AFDC (%)
AFDC
Payments ($)
AFDC Status
Recipient .608 --- --- --- - --
Applicant .392 -0.054*** -67.79 -0.005 +37.98
(0.020) (55.11) (0.020) (33.53)
Prior AFDC History
Received AFDC In 18 Months
Prior to Random Assignment
No Months .227 --- --- --- ---
I to 17 Months .301 -0.025 -94.15 +0.151*** .- 123.93 * **
(0.027) (73.53) (0.026) (44.74)
All 18 Months .472 -0.011 -112.16 ^0.202*** -321.72***(0.040) (107.65) (0.038) (65.50)
On AFDC for at Least 5 .517 -0.034* -182.19*** +0.061*** +85.92***Yeors (0.017) (46.69) (0.017) (28.41)
Average AFDC Payments in 5.82 -0.004 -8.15 +0.017*** +107.80***18 Months Prior to Random (0.004) ( 9.72) (0.003) (5.92)Assignment (in Thousands)
Prior Employment History
Ever Employed in the Quar- .268 +0.229*** +282.55*** -0.022 -88.37**ter Prior to Random (0.027) (73.38) (0.026) (44.65)Assignment
Ever Employed in the Year .394 +0.079*** +138.25** -0.028 -76.98'Prior to Random Assignment (0.024) (64.61) (0.023) (39.31)
Earnings Greater than .182 +0.116*** +487.25*** +0.026 +23.15$3000 in the Year Prior to (0.026) (70.16) (0.025) (42.69)Random Assignment
Unadjusted R2
0.159 0.112 0.173 0.283
Constant 0.128 12.42 0.410 626.45
Dependent Variable Mean 0.296 514.25 0.733 1095.68
Sample Size 3211 3211 3211 3211
(continued)
2;(3,-265-
TABLE F.1 (continued)
SOURCE: MDRC calculations from the County of San Diego AFDC records and the State of CaliforniaUnemployment insurance earnings records.
NOTES:
1986.
The sample for this table in lutes individuals who registered between July 1985 and June
Coefficients are estimated by ordinary least squares. Numbers in parentheses ore estimatedstandard errors.
'Employed' and 'Received AFDC' are dichotomous dummy variables. 'Earnings' and 'AFDCPayments' are dollar variables and include cases with zeta values for those not employed and for those notreceiving welfare.
A two-tail ed t-test was applied to each caef f iclent. Statistical significance level s areindicated as: *= 10 percent; ** = 5 percent; * = 1 percent.
266
TABLE F.2
SINIM
AFDC: EARNINGS AMONG EMPLOYED EXPERIMENTALS AND CONTROLS
Experimentols Controls Difference
Average Total Earnings,If Ever EmployedQuarters 2-4 ($) 3107.43 3257.22 -149.79
Average Total Earnings,If Ever Employed ($)Quorter of paidom Assignment 1064.40 1135.16 -70.75Quarter 2 1203.53 1376.45 -172.92Quarter 3 1312.24 1550.22 -37.98Quarter 4 1711.)2 1771.43 -60.61
Totol Number of EmployedExperimentols and Controls,Quarters 2-4 745 584
Number of Employed Experimen-tals and ControlsQuorter of Random Assignment 448 403Quarter 2 495 394Quarter 3 529 406Quarter 4 538 4'2
SOURCE: AORC calculations from the County of Son Di jo AFDC recordsand the State of California Unemployment Insurance earnings records.
`TOTES: The som;le for tnis table includes individuals who registeredbetween July 1985 and June 1986.
These data are calculated from the regression-adjustednumbers In Table 8.1 where the overoge earnings for o porticulor time periodare divided by the employment rote for the some period. This is onon - experimental comparison since the characteristics of emayedexperimentals differ from those of employed controls.
For oil meosuros, the quarter of random assignment refers tothe calendar quarter in which random assignment occurred. Because Quarter1, the quarter of random assignment, may contain some earnings and AFDCpayments from the period prior to random assignment, It is excluded from thesummory measures of follow-up.
TABLE F.3
SWIM
AFDC: SHORT -TERM IMPACTS ON MEASURED INCOME
Outcome andFollow-Up Period Experimentols Controls Difference
Total Average Measure)Income, Quarters 2-4 ($) 4879>18 49V.77 -22.59
Average Measured Income ($)OQuarter of Random Assignment 1489.94 1479.36 +10.59
Quarter 2 1657.17 1670.18 -13.01
Quarter 3 1617.15 1616.59 ' +0.56
Quarter 4 1604.86 1615.00 -10.15
r
Sample Size 1606 1605 3211
SOURCE: MDRC calculations from the County of Son Diego AFDC recordsand the State of California Unemployment Insurance earnings records.
NOTES: The somple for this table Includes individuals who enteredthe program between July 1985 and June 1986.
These data include zero values for sample members notemployed and for sample members not receiving welfare. Tnese data oreregression-adjusted using ordinary least squares, controlling for pre-randomassignment characteristics of sample members. There may be some
discrepancies in calculating sums and differences due to rounding.
For oil measures, the quarter of random assignment refers to
the calendar quarter in which random assignment occurred. Because Quarter
1, the quarter of random assignment, may contain some earnings and AFDC
payments from the period prior to random assignment, it is excluded from the
summary measures of follow-up.
A two-tolled t-test was applied to differences betweenexperimental and control groups. Statistical significance levels ore
Indicated as: = 10 percent; = 5 percenk; = 1 percent.
Merage measured income is defined as personal earnings plus
welfare payments received during a quarter.
3 '2
-268-
TABLE F.4
SWIM
AFOC EARLIER ANO LATER COHORTS: SHORT-TERM IMPACTS ON EMPLOYMENT.EARNINGS. WELFARE RECEIPT. AF) WELFARE PAYMENTS
Outcome arid Foilow-Up Period
AFOC Earlier Cohort I AFOC Later Cohort
Experimentais Controls Oifference Experimentois Controls Oifference
Ever Employed. Quarters 2-4 (x) 48.7 39.2 +9.5*** I 43.6 32.9 +10.8***
Average Number of Quarters withEmpl oyment. Quarters 2-4 1.03 0.81 +0.22*** I 0.91 0.69 +0.22***
Ever Employed (%)Quarter of Random Assignment 29.0 26.8 +2.2 26.5 23.1 +3.4*Quarter 2 32.6 25.6 +7.0*** 28.8 23.3 +5.5***Quarter 3 33.9 27.1 +6.8*** 31.8 23.1 +8.7***Quarter 4 36.5 28.3 +8.2*** 30.1 22.6 +7.5***
Average Total Earnings.Quarters 2-4 (S) 1512.11 1258.42 +253.70** 1362.92 1092.38 +270.54*
Average Total Earnings (M)Quarter of Random Assignment 315.99 301.93 +14.05 276.32 262.09 +14.23Quarter 2 392.50 350.02 +42.48 347.21 322.11 +25.10Quarter 3 496.05 417.12 +78.93* 500.30 360.91 +139.39**Quarter 4 623.57 491.28 +132.29** 515.41 409.36 +106.05*
Ever Received Any AFOC Payments,Quarters 2-5 (%) 90.3 90.7 -0.4 92.1 93.5 -1.3
Average Number of Months ReceivingAFOC Payments. Quarters 2-5 8.35 8.87 -0.52** I 8.88 9.42 -0.55***
Ever Received Any AFOC Payments (%)Quarter of Random Assignment 91.1 90.1 +1.0 91.3 92.9 -1.7Quarter 2 88.6 88.4 +0.1 90.9 91.5 -0.6Quarter 3 77.2 79.2 -2.1 81.0 84.4 -3.54Quarter 4 69.0 73.3 -4.3** 72.3 79.3 -7.0***Quarter 5 63.5 70.6 -7.0*** 68.7 74.6 -5.9*,?*
Average Total AFOC PaymentsReceived, Quarters 2-5 ($) 4258.61 4649.50 -390.89*** 4620.67 5044.05 -423.38***
Average AFOC Payments Received ($)flouter of Random Assignment 1197.57 1169.10 +28.48 1189.36 1223.72 -34.36Quarter 2 1253.29 1298.42 -45.12* 1325.39 1372.87 -47.48*Quarter 3 1072.21 1168.28 -96.07*** 1175.19 1293.85 -118.66***Quarter 4 908.82 1103.22 -114.40101* 1082.63 1228.70 -146.07***Quarter S 944.28 1079.58 -135.30**4 1037.46 1148.63 -111.17***
Sample Size 870 888 1758 732 721 1453
(continued)
330
TABLE F.4 (continued)
SOURCE: ANC calculations from the County of Son Diego AFDC records and the State of California Unemployment Insurance
earnings records.
NOTES: The earlier cohort registered between July 1985 and December 1985 and the later cohort registered between
January 1986 and June 1986.
These data include zero values for sample members not employed and for sample members not receiving welfare.
These data ore regression-adlusted using ordinary least squares. controlling for pre-random assignment characteristics of
sample members. There may be sane discrepancies in calculating suns and differences due to rounding.
For cii measures. the quarter of random assignment refers to the calendar quarter In which andom assignment
occurred. Because Quarter 1, the quarter of random assignment. may contain sane earnings and AFDC payments from the period
prior to random assignment. it Is excluded from the summary measures of follow -up.
A Iwo-tailed t-test was applied to differences between experimental and control groups. Statistical
significance levels ore indicated as: * = 10 percent: * = 5 percent; ** = 1 percent.
331
TABLE F.5
SWIM
AFDC-U: ESTIMATED REGRESSION COEFFICIENTS FOR
EMPLOYMENT AND WELFARE MEASURES IN QUARTER FOUR
VariableVariable
Mean
Ever
Employed (%) Earnings (8)
Received
AFDC (%)
AFDC
Payments (8)
Experimentol Group Member .512 +0.060** +99.93 -0.027 -113.62**
(0.025) (82.86) (0.024) (49.55)
Office
Son Diego West .493 --- --- --- - --
Service Center .507 +0.002 +175.15** -0.035 -42.33
(0.027) (88.67) (0.026) (53.13)
Age Greater Than or .397 -0.046 -252.22*** -0.009 -0.86Equal to 35 (0.029) (96.65) (0.028) (57.90)
Female .090 -0.019 -10.85 -0.016 -99.10
(0.044) (147.42) (0.042) (88.16)
High School DiPi0M0 or GED .470 +0.058** +117.14 +0.013 -51.38
(0.028) 02.13) (0.026) (55.07)
Marital Status
Never Married; Divorced; .147 ---- ---- ---- - - --
Widowed; Married, not
Living with Spouse
Married, Living with Spouse .853 +0.044 +130.38 -0.053 -59.97
(0.036) (121.2,', (0.035) (72.65)
Family Status
Any Children Less Than 6 .715 +0.012 +147.58 -0.037 +15J.67 **
(0.032) (106.47) (0.031) (63.55)
Any Children 6 to 18 .585 +0.005 +168.14 -0.048 +68.33(0.031) (103.07) (0.030) (61.43)
Rote /Ethnicity
White, Non - Hispanic .389 --- -__ ___ - --
Block, Non-Hispanic .208 +0.047 -37.34 +0.054 :1;;.i3*
(0.035) (116.42) (0.034) (6r,.77)
Hispanic .403 +0.044 +50.64 +0.071** +15.65
(0.031) (103.99) (0.030) (62.29)
(continued)
-271-334
TABLE F.5 (continued)
Variable
Voriable
Mean
Ever
Employed (%) Earnings ($)
Received
AFOC (s)
AFOC
Poyments ($)
AFOC Status
Recipient .405 --- --- --- - --
Applicant .595 -0.043 +1.55 -0.039 -119.77*
(0.035) (115.45) (0.033) (68.98)
Prior AFOC History
Received AFDC in 18 Month
Prior to Random Assignment
No Months .336 --- --- --- - --
1 to 17 Months .341 -0.045 -215.87 +0.183 +46.89
(0.040) (132.77) (0.038) (78.91)
All 18 Months .323 +0.000 -179.19 +0.196 -140.49
(0.069) (230.87) (0.066) (138.08)
On AFDC for at Least 5 .148 -0.018 -101.15 +0.051 +179.77*Years (0.039) (130.66) (0.038) (78 31)
Average AFOC Payments in 5.50 -0.008 -0.01 +0.018*** 90.3818 tonths Prior to Random (0.005) (0.02) (0.005) (10.80)
Assigment (in Thousands)
Prior Employment History
Ever Employed in the .382 +0.125* -90.07 -0.088** -71.60
Quarter Prior to Random (0.037) (143.22) (0.035) (73.50)
Assigraent
Ever Employed in the Year .565 +0.080 +42.90 +0.034 -50.16
Prior to Random Assignment (0.038) (129.42) (0.036) (74.72)
Earnings Greater than .327 +0.131*** -257.16* +0.012 -50.69
$3,000 in the fear Prior to (0.036) (149.73) (0.035) (;1.78)Random kssigment
Average Earnings In Year 3590.19 --- +0.02 --- - --
Prior to Random Assigment (0.02)
Average Earnings quarter 874.34 --- +0.24** --- - --
Prior to Ronan sgnment (0.05)
Average Earning: in Fourth 843.03 --- +0.20* --- - --
Quarter Prior to mom (0.05)
Assignment
(continued)
3:13
-272-
TABLE F.5 (continued)
VariableVariable
Mean
Ever
Employed (%) Earnings ($)ReceivedAFDC (%)
AFDC
Payments (s)
Unadjusted R2 0.126 .198 0.179 0.282
Constant 0.197 244.87 0.555 856.01
Dependent Variable Aeon 0.359 822.45 0.660 1221.08
Semple Size 1341 1341 1341 1341
SOURCE: See Table F.1.
NOTES: The sample for this table includes individuals who registered between July 1985 and June 1986.
Coefficients are estimated by ordinary least Fquares. Numbers in parentheses are estimatedstandard errors.
'Employed' and 'Received AFDC' are dichotomous dummy variables. 'Earnings' and 'AFDC Payments'are dollar variables and include cases with zero values for those not employed and for those not receivingwelfare.
The regression equations for earrings include three additional indicators of prior earnings. Adifferent model was used for earnings because the difference between experimental s and controls in priorearnings was statistically significant, prior earnings are highly correlated with future earnin9s, andconsequently, earnings impacts are affected. On the other hand. prior earnings were not included i n themodels for employment and welfare outcomes because prior earnings are less related to these outcomes and theimpacts are not affected in a meaningful manner.
A two - tailed t-test was applied to each coefficient. Stcti sti cal significance I evel s areindicated as: *= 10 percent; ** = 5 percent; ** = 1 percent.
3:14273
TABLE F.6
SWIM
AFDC -U: SNORT-TERM IMPACTS ON MEASURED INCOME
Outcome andFollow-Up Period Experimentals Controls Difference
Total Average Measured($)aIncome. Quarters 2-4 ($) 6146.66 6092.00 +54.66
Average Measured Income Mc'Quarter of Random Assignment 1864.88 1875.43 -10.56
Quarter 2 2077.92 2031.63 +46.29
Quarter 3 207,1.60 2013.29 +15.32
Quarter 4 2040.14 2047.08 -6.94
Somple Size 687 654 1341
SOURCE AND NOTES: See Table F.3.
TABLE F.7
9NIM
AFOC-U EARLIER AND LATER COHORTS: SHORT-TERM IMPACTS ON EAHLOYMENT,EARNINGS, WELFARE RECEIPT, AND WELFARE PAYMENTS
Outcome and Follow-Up Period
AFOC-U Earlier Cohort AFOC-U Later Cohort
Experlmentals Controls Oifferencs Experimentals Controls DifferenceEver Employed, Quarters 2-4 (%) 53.4 44.0 +9.341, 53.9 43.8 +10.0***Average limber of Quarters withEmployment. Quarters 2-4 1.16 0.92 +0.24*** 1.09 0.96 +0.13Ever Employed (%)Quarter of Random Assignment 38.5 36.7 +1.8 37.2 34.9 +2.3Quarter 2 36.8 27.1 +9.7.4 35.8 31.6 +4.2Quarter 3 38.6 31.9 +6.8 es 36.8 31.8 +5.0Quarter 4 40.6 32.9 +7.6* 36.7 32.8 +3.9
Average Total Earnings.Quarters 2-4 (s) 2303.41 2067.01 +236.40 2411.13 2010.32 +400.81
Average Total Earnings (5)Quarter of Random Assignment 584.74 673.47 -88.73 626.96 512.39 +114.57Quarter 2 654.88 585.96 +68.92 649.01 548.90 +100.10Quarter 3 806.77 727.12 +19.04 862.36 662.96 +199.40Quarter 4 841.76 753.33 +88.43 899.76 798.46 +101.31
Ever Received Any AFOC Payments.Quarters 2-5 (%) 85.3 85.3 +0.0 86.9 87.9 -1.0Average Number of Months ReceivingAFDC Payments, Quarters 2-5 7.44 7.99 -0.55' 7.68 7.90 -0.22Ever Received Any AFOC Payments (%)Quarter of Random Assignment 85.3 82.2 +3.2 86.1 86.7 -0.6Quarter 2 02.3 83.2 -0.9 84.3 84.9 -0.6quarter 3 65.3 71.5 -6.2* 69.7 70.7 -O./Quarter 4 63.4 68.2 -4.8 65.9 66.7 -0.8Quarter 5 59.9 63.7 -3.8 59.9 61.8 -1.9
Average Total AFOC PaymentsReceived, Quarters 2-5 (5) 4730.96 5351.72 -620.76*** 5009.24 5276.11 -266.87Average AFOC Payments Received (5)Quarter of Random Azslipment 1263.92 1252.19 +11.74 1260.53 1302.78 -42.24Quarter 2
1399.71 1468.90 -69.19 1427.52 1486.25 -58.73Quarter 3 1142.75 1328.12 -185.77*** 1236.57 1326.72 -90.16Quarter 41124.01 1307.92 -183.91"S 1210.96 1251.74 -40.78Quarter 5 1064.49 1246.38 -181.8944 1134.19 1211.39 -77.20
Sample Size375 348 723 312 306 618
SOURCE AND NOTES: See Table F.4.
336
FOOTNOTES
FOOTNOTES - CHAPTER 1
1. Several components of the SWIM program continued to operatethrough September 1987.
2. In this report, AFDC (called AFDC-FG in California) refers towelfare cases headed by a single parent. AFDC-U (calledAFDC-UP in California) refers to two-parent households wherethe principal earner is unemployed; all principal earners musthave had some connection to the labor force during the 12months prior to welfare application. The majority of AFDC-Ucases are headed by married men; the -leads of AFDC cases aremostly women. When the term welfare is used in this report,it refers to both the AFDC and AFDC -U programs.
3. A little more than one-third of all welfare adult applicantsand recipients are required to register foi work or trainingin WIN as a condition of receiving AFDC or AFDC-U benefits,i.e., arcs 'WIN-mandatory. 8 Heads of two-parent householdscovered by the program for unemployed parents (AFDC-U) areautomatically considered mandatory. Heads of single-parenthouseholds covered by the AFDC program are mandatory, unlessexempted because they are under 16 or o'ier 65 years of age,under 21 and enrolled full-time in school, sick or
incapacitated, the mother of a child under age 6, living in aremote area, a caretaker of a sick p:Nrson, or the spouse of aWIN registrant. Failure to register with WIN or toparticipate in program activities can lead to a sanction thatreduce. or eliminates the family's welfare grant.
4. For a more detailed discussion of pre-OBRA welfareinitiatives, see Goldman et al., pp. 4-22, 1984.
5. GAO, 1987.
6. Programs in eleven states were included in the demonstration,but only eight of the evaluations examined program effective-ness. See Gueron, 1987, for a summary of the demonstrationresults.
7. Friedlander et al., 1986.
8. Examples include the Employment and Training (ET) Choicesprogram in Massachusetts, the Realizing Economic Achievement(REACH) program in New Jersey, Project Chance in Illinois, andthe Greater Avenues for Independence (GAIN) program in'California. As described later in Chapter 1, GAIN was
-278- 3 :18
devel,',ded independently of SWIM
9. A key element in the bill was to shift responsibility foremployable recipients from ASS to EDD, whereby EDD would issuethe welfare checks. The intent was to form a closer tiebetween welfare and work by having the department in charge ofemployment issue the grant check. Because of EDD's concernabout the feasibility of this, EDD takeover of grant paymentsto employables was put off until a second phase. Also, pilotprojects were to be implemented in two counties to begin with;a third was added later.
10. See Goldman et al., 1985a, pp. 2-7 for a detailed discussionof the historical developments leading up to the EPP programin California and EWEP in San Diego.
11. The maximum number of work hours could not exceed the amountof the grant divided by the federal minimum wage, with thefurther restriction that participants be allowed one day aweek for individual job search.
12. For full results, see three MDRC reports on San Diego's EPP/EWEP program: Goldman et al., 1986; Goldman et al., 1985a;Goldman et al., 1984.
13. Note that gains for AFDC's in the group eligible only for jobsearch varied by cohort (or time of welfare application).
14. In San Diego, California's change to WIN Demonstration statusresulted in several changes to the EPP/EWEP model. Pro-grammatIcally, the most important changes prompted by WINDemonstration status were the following: Responsibility forWIN registration shifted fran EDD to the county welfare depart-ment; responsibility for initiating adjudication proceedingsfor registrants not complying with non-EDD activities was alsoshifted fran EDD to county staff; individuals enrolled andparticipating in self-initiated education programs wereallowed to be deferred from the program; and looser deferralcriteria were instituted for individuals in self-initiatedtraining programs.
15. Note that non -SWIM offices began to serve recipients as wellas applicants in the EPP/EWEP program in 1985.
16. Original plans called for implementation of the program earlyin 1985. Several factors, however, led to a decision by thecounty to delay program implementation until July 1, 1985.First, the original grant proposal contained insufficientfunds to support the level of research desired by SSA. Thecounty decided to delay program implt-nentation to allow aredirection of savings to the research effort. Second, San
-279-
Diego implemented the WIN DemoDstration program on July 1,1985. To change from regular WIN to the WIN Demonstrationprogram after SWIM start-up would have disrupted the SWIMproject. Third, the San Diego EDD district could not projectthe amount of money they would have available for the programuntil the end of their fiscal year. Fourth, the legislationpermitting the county to operate EWEi, was scheduled to expireJune 30, 1985 without assurance of continued legislativeauthority or funding.
17. AFDC-U families, as well as the case head, generally lose alltenefits when the head of the household is sanctioned. Whenan AFDC registrant is sanctioned, benefits are reduced by onlythe registrant's portion of the grant. The legislationspecified that families of AFDC-U registrants who weresanctioned for non - compliance in connection with EWEP shouldnot lose their AFDC benefits. To comply with thisrequirement, state aid was used to continue benefits for thefamilies of AFDC-U registrants sanctioned in connection withEWEP.
18. Three tiers of wage rates were used. As of September 1985,the wage rate tor individuals with no high school diploma was$4.19. The rate for those with a high school diploma but nohigher degree was $5.31. The rate for those with a collegedegree was $7.06.
19. Broken down by fiscal year, non-research demonstration moniesamountki to approximately $233,000 for October 1984 throughSeptember 1985; $742,000 for October 1985 through September:'f:86; and $725,000 for October 1986 through September 1987.Note that these monies extended beyond the 24-month pertodcovered in the evaluation.
20. The statistics in this section come fran several sources.Population numbers are from the California Department ofFinance. Percentages of county residents employed in variouslabor sectors are from the California Employment DevelopmentDepartment. Unemployment rates are fran the U.S. Bt.reau ofLabor Statistics. Percentages of county population living inpoverty, living in rural areas, having a high school diploma,and not u.ng English as a primary language are from the 1980Census.
21. This information comes from a one-page research interviewdocument administered at initial registration. Note that notall of this activity was approved by program staff as meetingprogram stand&rds t-Jncerning content, duration end credithos. According to SWIM automated tracking system data, 7.3percent of the AFDC registrants and 4.1 percent of the AFDC-Uregistrants were verified as inittally active in approved
-280-
340
self-initiated education or training activities.
22. Unemployment rates for the State of California and the UritedStates were, respectively, 7.8 and 7.5 percent in 1984; 7.2and 7.2 percent in 1985; 6.7 and 7.0 percent for 1986; and 5.6and 6.1 percent for May 1987.
23. In calculating the amount of each monthly assistance payment,not every dollar a recipient earns is subtracted from his/herwelfare check. As specified by the (federal) DeficitReduction Act of 1984 (DEFRA), the following disregards areapplied to earnings: First, to take work expenses intoaccount, a fixed amount ($75) is subtracted from the earnings.Then, up to $160 per child for the actual cost of childcare isdeducted. Finally, as a work incentive, for four months anadditional $30, plus one-third of the remainder of earnedincome not already disregarded, is deducted from earnings.For the next eight months, $30 is deducted.
As would be expected, a state with a high payment standardallows a greater proportion of the welfare population toreceive assistance while working, and the working recipientsmay have higher overall levels of earnings than those in lowgrant states.
According to an MDRC study of the relationship betweenearnings and welfare benefits for working recipients, SanDiego had a high proportion of welfare applicants who combinedwork and welfare in at least one month during a 12-monthfollow-up period. See Goldman et al., 1985b.
24. Note that at the end of the 12-month follow -up period, overone-fifth of those still registered with SWIM were employedand approximately half of the individuals who were stillactive were participating in an education or training program.
FOOTNOTES - CHAPTER 2
1. The components offered by the SWIM program evolved over time.STAR, a two-week employment search workshop, consisting oftraining in job search techniques and employment searchactivities in the field, replaced the bi-weekly Job Clubcomponent (associated with EWEP) in January 1987. STAR, a
more intensive component, could not be done concurrently withEWEP. Therefore, those who were eligible for PREP werereferred to STAR first. After completion of STAR, registramswere referred to EWEP. The implementation of GAIN affectedSWIM in several ways. Late in 1986, registrants began to takea literacy test as part of the orientation process. This test
-281-
341Mil
was used as a pilot for GAIN. The implementation of GAIN inSan Diego also affected the types of activities that wereavailable during May and June 1987. In aaticipation of a July1, 1987 GAIN implementation date, many SWIM activities wereeither phased out or became short -term in nature during thesemonths. For example, EWEP assignments made during thesemonths were scheduled to end June 30, 1987. Long-termtraining assignments which utilizd ;a,b payment of training-related expenses were not made during the last few months ofSWIM since, according to the schedule, this type of fundingwould not be available after June 1987.
2. Interviews with supervisors from non-SWIM county officesindicated that the following differences existed between SWIMand the general county program: First, tracking of
registrants' activities was not as extensive in the non-SWIMoffices as it was in the SWIM offices. Second, in non-SWIMoffices, social workers, not CRU (Coordination and ReferralUnit) staf, would conduct assessments. Third, althoughnon-SWIM offices could refer registrants to communitIducation or training programs after ENEP or job club,non-SWIM office registrants were not eligible for ISESA.F 'rth, staff in non-SWIM office_ did little follow-up onregistrants referrgd to education or training programs.Fifth, follow-up on employed registrants or those inself-initiated education or training did not occur asfrequently as it did in the non-SWIM offices. Employment wasgenerally verified at 30 days after, the employment began andthen every 6 months or year afterwards. Self-initiatedregistrants were generally asked to verify school enrollmentat tht beginning of each semester by providing school forms orsigning statements at the program office verifying theirattendance.
3. Sixty-two of these registrants were excluded from the impactanalysis because they had no social security number. Socialsecurity numbers were used to access earnings records. Theseregistrants were included in all other analyseo.
4. Note that in addition to demographic differences between thesetwo assistance categories, different procedures govern thecalculation of welfzre grants for the two assistance groups.
5. AFDC SWIM- eligibles who registered betweenJune 31, 1986, were weighted by a factorAFDC-U SWIM-eligibles alio registered betweenJune 30, 1986, were weighted by a factorAFDC SWIM - eligibles who registered betweenJune 30, 1987, were weighted by a factorAFDC SWIM-eligibles who registered betweenJune 30, 1987 were weighted by a factor of 3
-282-
July 1, 1985, andof 1.8571428571.July 1, 1985, andof 2.3333333333.
July 1, 1986, andof 2.8571428571.
July 1, 1986, and.3333333333.
6. During the second year of SWIM a few current WIN-mandatorieswho transferred into the SWIM offices from non-SWIM officeswere taken into the sample.
7. The CIS completion rate was very high. A quality check ofresponses to the important demographic questions on the CISrevealed only 5 registrants in the impact sample were missingresponses to any of these questions. Of these fiveregistrants, 4 were missing only one response to the questicnschecked. One registrant was missing two responses.
8. Earnings reported for the third quarter may be preliminaryestimates, since some adjustments to earnings may occur as aresult of future reporting by employers.
9. UI earnings data were compared to previous employment recordedon the CIS forms of those individuals who reported having beenemployed for 19 or more months in the two years before randomassignment. UI-reported earnings in the year before randomassignment were found for 81 percent of the people whoreported employment on the CIS.
10. Welfare payment records during the 20 months prior to randomassignment were found for over 92 percent of sample memberswho reported having had their own AFDC case for more than twoyears before random assignment according to the CIS.
11. The SWIM Automated Tracking System data only capturesinformation known to the program. It does not captureactivities in community service programs or employment unlessthey were known to program staff.
FOOTNOTES - CHAPTER 3
1. This number represents the equivalent of 18 full-time staffmembers. In total, 19 staff were involved in SWIM, two ofwhom were part time.
2. SWIM clerks monitored the continuation of employment in twoways. The most common method involved registrant verificationof their own employment. Every days, clerks mailed formsto registrants which requested their employer's name, address,telephone, employment sta;:t date, work hours and rate of pay.The registrant would sign the form, certifying that he/she wasstill employed, and mail the form back to the clerk. Clerksalso verified employmen': by checking whether income was usedto calculate a registrant's welfare grant. An inherent lamexisted in this procedure, however, since the income included
-283-
343
in a current grant calculation was what the registrant hadearned two months earlier.
If verification forms were not received in a timely manner,returned forms indicated that the registrant had stoppedworking, or welfare records indicated a lack of earned income,the clerk notified a SWIM JDC. The JDC would mail a letter tothe registrant requesting that he/she cane into the office foran appraisal. If the registrant did not attend the appraisal,another would be scheduled. If the second appraisal wasmissed, the case would be referred to a social worker to startthe determination/sanctioning process.
3. June 1986 was the last month in which a proportion of new SWIMregistrants were assigned to a control group. Starting inJuly 1986, the only research-related responsibility of theSWIM clerks was checking that new registrants had not beenpreviously assigned to the control group.
4. Caseload numbers were obtained through interviews with programstaff and verified through MDRC calculations using data fromthe SWIM automated tracking system.
FOOTNOTES - CHAPTER 4
1. Sanctioned individuals were also required to re-register withthe program once their sanctioning r.tiod was over.
2. Following the interview, SWIM staff were notified as to who toexpect for registration/orieatation. These applicants andre-determined recipients were told to contact a program officewithin one day to schedule a registration appointment. At theSWIM offices, receptionists took the telephone calls andscheduled individuals for registration/orientation within fiveworking days. Clients were expected to bring signed registra-tion forms back to their IM worker as a cond.' 'on of welfareeligibility. Note that 'automatic? registra .on procedureswere not in effect during the time period that SWIM operated.Individuals were not considered to be registered until theycompleted the appropriate forms at program offices.
3. Although approximately one-third of the 'renewal' recipientswho were sent appointment notices through the mail did notattend their initial registration appointment, interviews withstaff indicated that nearly all of those who were sentfollow-up nctices eventually registered. Data concerning theproportion of apilical,%, or re-determined recipients whofailed to register are not available.
-284-
344
4. Registration/orientation sessions were held five mornings perweek at one SWIM office; four mornings per week at the other.
5. The research status of all individuals who came to the SWIMoffices for registration was investigated before theregistration/orientation session began. Any control groupmembers among the individuals awaiting registration would bereferred to a social worker, who would help them completetheir registration forms and inquire about their socialservice needs. After this short meeting (about 5 minutes),these individuals could leave.
6. In one office, this was the first step in the registra-tion/orientation process. In the other office, registrantswould first meet as a group for a 5-minute overview of thescheduled registration and orientation activities.
7. In one of the SWIM offices, individuals were randomly assignedto a r)ntrol or SWIM-eligib]e group before orientation tookplace. In this office, control grout) members and thoseeligible for the program attended Lparate orientationsessions. In the control group orientation, individuals weregiven information about available program social services(such as the $50 entered employment stipends) and communityresources, and an explanation of their program rights andresponsibilities. The orientation for those eligible for SWIMpresented more information about the program.
In the other SWIM office, random assignment took place whileindividuals were attending the orientation session. In thisoffice, control group members and those eligible for theprogram remained together for orientation. During orient-ation, registrants were told that only some of them would gothrough the program. At the close of orientation, the sessionleader was informed as to who had been assigned to the controlgroup. After releasing the controls from the orientationsession, staff told the rest of the group that they had beenselected for the program.
C. Starting in October 1986, the administration of an hour-longliteracy test was included in the registration/orientationprocess. The test results were used for GAIN planningpurposes.
9. Control group members did, however, remain in contact withtheir eligibility technicians.
10. Individuals referred to EDD staff for assignment wouldcomplete several forms and be given more detailed informationabout the job search workshops.
-285-
345
11. Undocumented workers who were the parents of United Statescitizen children could receive welfare benefits on behalf oftheir children. As a condition of welfare eligibility, theseundocumented workers were required to register with SWIM,following usual WIN-mandatory regulations. However, they werenot required to participate in the program because of theirinability to work legally in this country.
12. In the second office sane job club leaders were !dismissing'registrants from the component once they had attended sixsessions.
13. The JDU job developers had a fairly extensive 'bank' of jobs.Same of these job orders were procured as OJT slots. However,staff acknowledged that these positions rarely involved anyskills training; staff referred to OJT as 'buying a job for aclient.' In looking for registrants to fill available jobopenings_ no distinctions were made between OJT positions andunsubsidized positions.
14. CRU staff were informed of problems only on an exceptionbasis, that is, if a referred registrant never attended orient-ation or dropped out of the program. If an attendance problemwas identified, CRU staff would schedule an office meetingwith the registrant. If the registrant missed two sequentialmeetings, the case would be referred to a social worker for adetermination interview. If a registrant failed to attend twointerviews with a social worker, a sanction would berequested.
15. For those who missed an orientation, the conciliation planspecified that they attend the next one scheduled. For thosewho had missed assigned days at a worksite, the conciliationplan specified two weeks of the next month's assignment inwhich the registrant had to work all assigned days or a
sanction would be automatically requestcl. If the registrantadhered to the conciliation plan, the determination processended. Later noncompliance would start the procet's overagain.
16. For several years the county had had an agreement with RETC tofund JDU activities. At the beginning of SWIM, this agreementwas restructured to include ISESA as a JDU activity. Nochanges to the contract, however, were made in relation totraining activities. The few new agreements drawn up at thebeginning of SWIM were done to avoid having SWIM registrantsdisplace regular continuing education students. As the SWIMprogram got underway these agreements were not necessary andoften not remembered, due to the relatively small numbers ofSWIM registrants referred to community resources.
-286-
346
17. Registrants were informed of their scheduled assessmentappointments via a SWIM appointment notice mailed to them atleast 10 days prior to the scheduled date. The appointmentletter did not specifically mention education or training; itsimply noted that the regis nt had been scheduled for aninterview to review/develop his/her employment plan anddiscuss the services that the program could provide. The formnoted on the bottom that failure to attend the interview couldresult in reduction or discontinuance of the registrant's AFDCgrant.
18. Registrants would fill out questionnaires regarding theiremployment and educational histories, take vocational teststhat measured interests, aptitudes and values, and then reviewthe test results individually with a counselor. Counselorswould make recommendations concerning registrants' suitabilityfor different types of work and suggest community collegeprograms which provided training for these occupations.
19. For the first fail%re to attend, registrants were automatical-ly rescheduled (with a notice) for another assessment appoint-ment. After the second no-show (some CRU workers allowedthree missed appointments), the situation generally wasreferred to a social worker. The social worker then scheduledthe registrant for a reappraisal interview and, if theregistrant again failed to come into the office, automaticallyscheduled a second reappraisal interview. Registrants failingto attend either of these interviews were sanctioned.
20. ESL programs generally required attendance for three hours perday, five days per week; ABE involved ten hours per week;vocational programs required attendance for six hours per day,five days per week,
21. Note that SWIM did not set a limit on how long registrantscould remain in education or training programs.
22. To refer an individual to this program, CRU staff firstidentified an individual as in need of ABE. Next, centralwelfare office staff reviewed the individual's score on aliteracy test given to all WIN-mandatory registrants in thecounty (after October 1986) in preparation for the GAINprogram. If the registrant's test score indicated that he/shewould be reqired to seek ABE under GAIN, the individual wasreferred to one of the centers. If not, CRU staff wereinformed that the registrant should be referred to anotherprovider.
23. Most commonly, a registrar at the school would certifyattendance by checking with the registrant's teacher orconsulting roll books; enrollment was not considered to be
-287-
347
synonymous with participation.
24. The primary reason that noncompliance in education and
training was not sanctionable in its own right concerned
registrants' eligibility for training-related expenses.
Program planners envisioned that extensive support service
payments would be required if education and training weredeemed mandatory. Thinking that the costs of these supportservices would have been prohibitive, especially if
rc istrants in self-initiated education or training wereeligible for these services, program regulations were writtenin such a way that sanctions could be imposed for failing toattend office appointments but not for failing to attendeducation or training programs.
25. If a registrant was in training and the training did not meetthis criterion, a deferral could be made if all the followingconditions were met: the course was to be completed within oneyear; the training led to probable employment in the locallabor market; and the training appeared necessary for theregistrant to became competitive in the labor market at asuitable wage level.
26. If a registrant failed to return a form, however, CRU workersscheduled an office appointment. If the registrant hadcompleted a program and was not employed, the registrant wasassigned to another SWIM component. If the registrant failedto attend the appointment, adjudication procedures identicalto those for program-referred registrants were followed.
FOOTNOTES - CHAPTER 5
1. Since participation rates in this section reflect data fromthe SWIM tracking system, these indicators show the extent ofparticipation in activities known to and approved by programstaff. Thus, registrant-initiated education or trainingactivities are included to the extent that they met programstandards as to content, intensity, or duration, and to theextent that program staff were aware of them. This definitionof activity parallels the one used both by programadministrators and by line staff.
2. Not all these individuals participated in job search workshopsas their first component. Some individuals initially parti-cipated in self-initiated education or training, or wereemployed as of initial registration and participated in aworkshop later in their program tenure.
3. As explained in Chapter 4, if program staff approved an
-288-
348
individual's self-initiated education or training program (adecision based on the content, intensity and duration of theprogram), the registrant would be deferred from assignment toa job search workshop or EWEP for as long as the registrantremained active in the education or training program.
4. This report defines sanction differently from the way it isdefined in MDRC's other work/welfare reports. In thosereports, an individual was assumed to be sanctioned if aderegistration occurred after program staff had requested thata sanction be imposed. In SWIM, sanctions were defined morerigorously. Program staff recorded sanctions in the SWIMtracking system only after receiving a notice from IM staffindicating that a sanction had been imposed and the individualhad been deregistered for this reason.
5. However, caution is required in drawing comparisons with evalu-ations conducted as part of the Work/Welfare Demonstration dueto differences in the point at which research gr'ups wereidentified. For example, individuals applying for AFDC duringthe EPP/EWEP evaluation were randomly assigned at application,not at program registration. In that evaluation, controlscould have been sanctioned for not registering with theprogram. In the SWIM evaluation, random assignment occurredat registration. Thus, once identified as a control, it wouldnot be possible for this individual to be sanctioned forfailure to register.
6. Sample members were categorized as applicants or recipientsbased on their status as of program registration.
7. The proportion of applicants who were denied welfare wasapproximated by examining the percent of applicants who didnot receive welfare Ciring the month in which they initiallyregistered with the program as well as the month followingthis registration. This figure was 14 percent for AFDCapplicants and 17 percent for AFDC-U applicants.
8. wo points should be noted concerning Table 5.4. First, thetable only covers the first and second ac,:ivites that occurredwithin 12 months of registration. Second, because participa-tion in job clubs was usually concurrent with participation inEWEP, job clubs were not included in Table 5.4. The 6 percentof the AFDC's and 4 percent of the AFDC-U's noted on Table 5.4as registrants who initially participated, in program-arrangededucation or training are probably individuals who initiallyparticipated in job clubs (without concurrent participation inEWEP) and whose second activity was program-arranged educationor training.
9. The percentage of individuals reported in this chapter as
-289-34D
active in self-initiated education or training as of initialprogram registration differs from the percentage reporte. inChapter 2 for several reasons. First, the figure cited inChapter 2 (based on a registrant questionnaire completed atinitial program registration) includes any type of educationor training. The education or training activities included inthe analyses in this chapter are only those which were!approved` by the program, meaning that these activities metprogram standards as to content, intensity, or duration.
Second, for sane individuals who initially reported that theywere active in education and training, schools could notverify their attendance. Unless attendance was verified,program staff did not record participation in the SWIMautomated tracking system, the source of data for this
chapter.
10. In this section, all averages calculated from enrollment datainclude weekend days.
11. Individuals who participated during the month but werederegistered during the same month were excluded from thesample. Also, the original sample consisted of 144registrants. Twenty-three individuals were dropped from theanalysis due to miscodes in the automated tracking system,computer sample selection problems, or incomplete case files.Finally, although all sampled registrants were AFDC's as c!initial registration, the case file reviews indicated that onesample member was an AFDC-U recipient during the review month.
12. The data base includes the child's age, the providers ofchildcare e ring SWIM activities, the location of the care,and whether the program paid for childcare during the month.Information on whether the childcare provider was licensed,the amount paid for childcare, and the funding source for paidchildcare was also collected, but was not of high enoughquality to analyze.
13. Note that one registrant was a 16-18 year-old child on his/herparent's case. This youth was designated the WIN-mandatoryregistrant of the case. The registrant, who had no children,was considered the 'child' of the case in this study.
14. One participant, whose youngest child was almost 14 years oldand considered old enough to care for him/herself, is includedin this statistic but not in the 22 percent figure given forparticipants whose children were all at least 14 years old.
-290-
350
FOOTNOTES - CHAPTER 6
1. About 13 percent of the nonparticipants did not receivewelfare during the month in which they initially registeredwith the program as well as the month following thisregistration. It is likely that rhese individuals wereeligible for the program for only a short time because theirwelfare applications were denied. Although almost 15 percentof the nonparticipants were eligible for the program duringthe whole 12-month follow-up period, it should be noted that 7percent of the nonparticipants were undocumented workers.These individuals did not qualify for program services but mayhave been registered for long periods of time.
2. Excluding nonparticipants from the analysis indicates theextent to which the program imposed an ongoing participationrequirement on those who participated at least one day. Theseresults, calculated only for participants, indicate that onlya relatively small proportion of all participants -- 21percent -- were active during all the months in which theywere eligible for program services. However, 48 percent ofall participants were active in almost all (at least 70percent) of their program-eligible months.
3. In this study, noncompliance was defined as fellows: Aregistrant was considered to be out of compliance with programrequirements whenever he/she did not participate in orcomplete an activity to which he/she was assigned. Note thatif a registrant was assigned to a job search workshop and didnot complete it (even if it was later discovered that he/shewas denied welfare, deregistered, or found a job), this wasconsidered noncompliance. Noncompliance also includedsituations where the registrant informed program staff of alegitimate reason for an absence prior to the scheduled startof an activity. Further, noncompliance included situationswhere program staff rescinded an individual's assignment priorto the scheduled start date of the activity. This occurred inten of the 242 cases reviewed. In these cases, assignmentswere rescinded because program staff discovered that theregistrant was pregnant, an undocumented worker, an excludedparent, no longer mandatory for other reasons, alreadyworking, or not on welfare. Note also that noncompliancerefers only to program requirements -- not other welfarerequirements. For example, refusal to apply for UI benefitsor failure to submit an income reporting form was notconsidered noncompliance in this study.
4. For registrants who did not comply with program requirementsduring the 15 to 18 month follow-up period, detailed infor-
-291-
3.51
mation was collected on the first program component in whichthey were noncompliant. Where applicable, informationconcerning the second instance of noncompliance was alsorecorded, although not in as much detail as the initialinstance. Although registrants could have been noncompliantin more than two program activities (e.g., the job searchworkshop, a reappraisal interview and EWEP), detailed infor-mation was not recorded for subsequent instances of noncom-pliance. As a result, the analysis presented in this chapterrepresents 90 percent of all instances of noncompliance notedfor the sample in case file folders. The results of the casefile reviews are presented according to the program activityin which the requirements were not met. Consequently, theaction or treatment of registrants noncompliant in PREP, forexample, are analyzed together, regardless of whether thisactivity represented the first or second instance of
noncompliance for the registrant.
5. The Notice of Action mailed to the registrant contained infor-mation concerning where to obtain free legal help. This formlisted a state toll-free hearing and legal aid informationnumber as well as the addresses and telephone numbers of locallegal aid offices and welfare rights groups.
6. In order not to have their welfare ben:fits cut while waitingfor a hearing, registra' s had to appeal the sanction withinten days. a registra.it did not appeal within ten days, the
sanction would be imposed. The registrant then had 90 days tocontest the action; if he/she won the appeal, the sanctionedamount of the welfare check was refunded to the registrant.
7. Monthly participation rates, described in detail in Chapter 7,are also problematic in this regard. Consistently high month-ly participation rates over the duration of a program do notnecessarily mean that the program reached every individual inthe targeted caseload or that a mandatory participation
requirement was enforced. First, the individuals counted asinactive in each month's rate might be the same individualseach month; this would indicate that a core group of individ-uals never participated. Second, if the group for which theserates are calculated consists of individuals still registeredwith and eligible for the program, sanctions are not takeninto account. (Individuals who are sanctioned arederegistered from the program during the .:anction period andtherefore are not calculated into monthly participation rateswhile the sanction is in effect.) Thus, monthly participationrates may mask the fact that sane individuals did not partici-pate in the program but were sanctioned for nonparticipation.
-292-
37
8. Note that employment was not counted as participation in thisanalysis. In addition, UI earnings records -- not programtracking data -- were used to estimate the proportion ofindividuals who held jobs during this period. These data didnot provide sufficient follow -up on all sample members.Consequently, the sample used to calculate the coveragestatistics comprised i'dividuals who registered between July1985 and April 1986.
9. Registrants generally participated within three months ofinitial program registration if they participated at all.Consequently, if coverage is measured at nine months afterregistration, ins%lad of at the 12-month point, the proportionof those who were not subject to the participation requirementis still low: At nine months after registration, only 4percent of both AFDC's and AFDC-U's had remained eligible forSWIM, were not employed, had never participated in anactivity, and had never been sanctioned. As of six monthsafter registration, the percent of the sample uncovered isalso low: 6 percent of both AFDC's and AFDC-U's.
10. As noted above, caution is required in drawing comparisonswith other programs evaluated as part of MDRC's Work/WelfareDemonstration due to differences in the point at whichresearch groups were identified, length of research follow-up,target population, etc.
FOOTNOTES - CHAPTER 7
1. In fact, most of the analysis in this chapter adopts a
relatively strict definition of the type of activity thatconstituted !program participation!: To be considered activein a month, an individual must have attended a program-approved activity (including employment) for at least one dayduring the month. On the other hand, in terms of howintensively individuals had to participate in order to becounted as participants, this definition is not very strict.Although it is beyond the scope of this report, it is possibleto define participants as those who attend a certain number ofactivity sessions or claeses.
2. Sample sizes used throughout this chapter are weighted ones,representing actual SWIM- eligible caseload sizes. See Chapter2 for a full explanation of the weighting factors.
3. As described in earlier chapters, registrants could be activein more than one type of component at one time. Additionally,in any given month, an individual could finish one type ofcomponent and begin another. In the monthly participation
-2 93-
3:1)3
rates calculated throughout this chapter, program activitieshave been prioritized in such a way that individuals arecounted as active in only one type of activity during eachmonth. In the rates, a higher priority was placed onactivities which are most often thought of as 'program
participation." Program-arranged activities were givenpriority over self-initiated activities as well as employment;also, a higher priority was placed on self-initiated activi-ties than on employment. Note that self-initiated activitiesincluded education, training and union or other self-initiatedjob search. Within the broad category of program-arrangedactivities, priorities were set in the following order: Workexperience activities (including EWEP and OJT) were givenhighest priority. Participation in job search activities(including job search workshops, job clubs or STAR, and ISESA)was next, followed by program-arranged education or training.Two examples can illustrate the effects of these priorities.A registrant who participated in job cub concurrent with EWEPwould hate been counted in the work experience category. Ifthis same registrant subsequently found a job during the samemonth he/she participated in EWER and job club, the registrantstill would have been counted in the work experience category.
For clarity of presentation, sc,me of the detailed breakdownsof types of activities have been collapsed in this chapter.For example, this chapter does not distinguish, as doesChapter 5, between education and training, or betw2eneducation/training provided by community colleges, JTPA orother providers.
4. In varying the definition of participation, the type of
activities counting towards participation is not the wilyfactor that can be examined. As noted above, it is alsopossible to investigate how changing the intensity of theparticipation requirement would affect monthly participationrates. Throughout this chapter, individuals who participatedin EWEP for at least one hour or any other activity for atleast one day during the month were categorized asparticipants.
5. Caution is required in drawing comparisons between theseparticipation rates and those calculated as part of MDRC'sDemonstration of Work/Welfare Initiatives, due to differentprogram settings, various data quality issues, measurementissues, different target populations, large or stall controlgroups, and various program models. However, it isinteresting to contrast these participation rates with thosecalculated for West Virginia's statewide unpaid workexperience program -- a straight-forward work program in whichthe assignment lasts as long as the recipient receiveswelfare. The state successfully imposed the requirement for
-294-
3S4
the AFDC-U registrants but did not impose it rigorously forAFDC registrants. For AFDC-U registrants, counting only workexperience as participation, the program achieved monthlyparticipation rates of between 59 and 69 percent of thecaseload on a monthly basis. These rates were calculated forMay 1983 through February 1984 -- a steady-state period of theprogram. For AFDC registrants, rates were calculated forOctober 1983 through April 1984, a period in which the programwas still phasing in its current WIN-mandatory caseload. AFDCmonthly participation rates were much lower (again, countingonly work experience), ranging from 13 to 21 percent on amonthly basis.
6. However, during the first few months of the demonstration,before Spanish-speaking job search workshops and EWEPworksites were developed, registrants who spoke only Spanishwere immediately referred to program-arranged ESL courses.Once other program components were developed for theseindividuals, they were referred to workshops and/or EWEPbefore any type of program-arranged education.
7. Mathematically, this differential rule has the followingeffect: /FDC-U registrants who are employed are taken out ofboth the numerator and the denominator of the monthlyparticipation rate. AFDC registrants who are employed remainin both the numerator and denominator of the rate.
8. The effect of varying the definition of program eligibles isthe same when only program-arranged activities, or program-arranged and self-initiated activities, are taken ,ntoaccount.
9. Mathematically, changing the definition of 'program- eligibles'in this way increases the denominator of the rate to a greaterextent than the numerator. For example, during November 1986,2,912 were registered throughout the month; 1,614, or 55.4percent, were active at any point during the month. In thesame month, 3,221 were registered as of the end of the month;1,703, or 52.9 percent of these individuals, were active atany point during the month (not necessarily at the end of themonth). Lastly, 3,468 were registered at least one day duringthe month (but not necessarily as of the end of the month);1,800, or 51.9 percent of these individuals, were active atany point during the month.
10. Also note that Figure 7.8 defines program-eligibles as thosewho were registered at least one day during the month orquarter. When eligibility is defined as those registered asof the end of each month or quarter, the same relationshipbetween monthly and quarterly participation rates exists.
-295-
3 5 5
11. This question is similar to one posed in (...aptet 6 in
reference to the extent to which the program imposed anongoing participation requiremen*. In Chapter 6, however,eligibility and participation patterns were analyzed from thepoint of view of the registrants. To do this, program
eligibility and participation were examined for each
registra't over a 12 -month follow-up period. The 12 monthswhich this period represented were not the same 12 calendarmonths for each registrant. This chapter examt.ms eligibilityand participation frrt a program operations point of view. Todo this, eligibility and participation are examine for eachcalendar month of the demonstration.
12. The methods used to define those who were inactive during theselected months (and thus select the sample) were slightlydifferent from those used earlier in the chapter. First, asmall number of individuals who were deregistered during theselected months were excluded from this analysis. Second, oneor two individuals were considered active in this analysisbecause they attended an EWEP orientation during the month.This was not considered participating in the rest o2 the
chapter. Third, th'. sample for this analysis was selectedfrc an early computer file. Several variables on this filewelt,: incorrect, but were corrected for the computer file usedfor the analysis in the rest of the chapter. Thus, althoughall individuals included in the analysis were nonparticipants,according to the SWIM automated tracking system, there is apossibility that the sample is not completely repr mtative.
13. Note that the sample included two individuals who werenonparticipants in bofth July and November 1986.
14. These individuals were not excluded from the analysis samplebecause they do shed some light on the status of those whowere counted as 'inactive' in the rest of the analysis.However, it is not likely that tee monthly participation ratespresented in the earlier sections ;'ere underestimates. Errorsin the oppo,f.to ditection are also probabl.i. That is, someregistrants recorded as participants in the SWIM automatedtracking system were probably, in fact, not active.
FOOTNOTES - CHAPTER 8
1. Bane and Ellwood, 1983.
2. Appendix Table F.1 presents the regression models used In thisanalysis.
3. This may be iue to the tact that recipients were phased into
-296-
the program at a slower pace than applicants. Differencesbetween early and later enrollees may also be partly due to animproving labor market. There was a decline in the unemploy-ment rate in San Diego county over the intake period for theimpact sample; the unemployment rate decreased from 5.3percent in the last six months of 1985 to 4.9 in the first sixmonths of 1986. The effect of the unemployment rate decliningis that the more advantaged remain employed or find jobs morequickly and therefore are less likely to apply for welfare.
4. For AFDC-U's, Appendix Table F.5 presents the regression modelused in this analysis.
5, For AFDC-U's, the regression equations for earnings (and out-comes that include earnings, such as measured income) use adifferent model from that used with the employment and welfareoutcomes. The model for earnings includes additional indicat-ors of prior earnings, that is, earnings in the year, quarterand fourth quarter prior to random assignment. As indicatedin Chapter 2, the difference between AFDC-U experimentals andcontrols in prior earnings were statistically significant.This, combined with the fact that prior earnings are highlycorrelated with future earnings, means that one should controlfor these prior differences when comparing the earnings ofexperimentals and controls. On the other hand, prior earningsare less related to the other outcome measures and controllingfor them did not lhange the experimental-control differenceson these outcomes in a meaningful manner, although impacts insome cases were slightly lower using the model with additionalearnings variables. For this report, the additional measuresof prior earnings are only included in the equations forearnings (and outcomes composed of earnings) and not for otheroutcomes. Alternative specifications of the model will beexplored ia greater depth for the final report.
6. When the sample is divided into subgroups, the sample sizesfor each analysis are considerably smaller, reducing thelikelihood of obtaining statistically significant effects.
7. Another difference between the AFDC-U and the AFDC compari-sons is that the percentage of applicants and recipients didnot differ between the two samples among the AFDC-U regis-trants as it did among the AFDC registrants. It is unclearwhy AFDC's and AFDC-U's would have been affected differently.
8. Goldman et al., 1986.
9. The SWIM offices were located in poor urban areas in thecounty in contrast to the other offices. This meant that thepopulation served in the SWIM offices were more disadvantagedand faced a more depressed local job market.
-297-
3
10. The sanctioning rules for the AFDC-U registrants also differedbetween the two demonstrations. Usually, the entire AFDC-Ucase is closed if the registrant refused to participate. In
SWIM, this rule applied to all activities except workexperience, in which case only the registrant would be
sanctioned if there were participation problems. In EPP/EWEP,this rule applied to all activities with no exceptions.
-298-
or-)
REFERENCES
Bane, M. J.; and Elwood, D. 1983. 'The Dynamics of Dependence: The Routesto Self-Sufficiency,' Cambridge, MA: Urban Systems Research andEngineering, Inc.
Friedlander, D.; Erickson, M.; Hamilton, G.; Knox, V.; with Goldman, B.;Gueron, J.; and Long, D. 1986. West Virginia: "anal Report on theCommunity Work Experience Demonstrations. New York: ManpowerDemonstration Research Corporation.
Got man, B.; Gueron, J.; Ball, J.; and Price, M. 1984. Preliminary Find-ings from the San Diego Job Search and Work Experience Demonstration.New York: Manpower Demonstration Research Corporation.
Goldman, B.; Friedlander, D.; Gueron, J.; Long, D.; with Hamilton, G.; andHoerz, G. 1985a. Findings From the San Diego Job Search and Workxperience Demonstration. New York: Manpower Demonstration Research
Corporation.
Goldman, B.; Cavin, E.; Erickson, M.; Hamilton, G.; Hasseibring, D.; andReynolds, S. 1985b. Relationship Between Earnings and WelfareBenefits for Working Recipients: Four Area Case Studies. New York:Manpower Demonstration Research Corporation.
Goldman, B.; Friedlander, D.; Long, D.; with Erickson, M.; and Gueron, J.1986. Final Report on the San Diego Job Search and Work Experience-'r')nstration. New York: Manpower Demonstration Research Corporation.
Gueron, J. 1987. Reforming Welfare With Work. Net York: The FordFoundation.
Subcommittee on Intergovernmental Relations and Human Resources, committeeon Government Operations, U.S. House of Representatives. 1987. Workand Welfare: Current AFDC Work Programs and Implications for Fee calPolicy. Washington, D.C.: General Accounting Office.
3 5-299-
PUBLISHED AND FORTHCOMING STUDIESIN THE MDRC DEMONSTRATXON OF STATE WORK/WELFARE INITIATIVES
MONOGRAPHS
Gueron, Judith. 1986.From a Multi-State
Work Initiatives for Welfare Recipients: LessonsExperiment.
Gueron, Judith. 1987.Ford Foundation.)
REPORTS
ARIZONA
Reforming Welfare with Work. (Published by the
Sherwood, Kay. 1984. Management Lessons From the Arizona WIN DemonstrationProgram.
ARKANSAS
Quint, Janet; with Goldman, Barbara; and Gueron, Judith. 1984. InterimFindings from the Arkansas WIN Demonstration Program.
Friw-ander, Daniel; Hoerz, Gregory; Quint, Janet; Riccio, James; withGoldman, Barbara; Gueron, Judith; and I. David. 1985. Arkansas:Final Report on the WORK Program in Two Counties.
Friedlander, Daniel; Goldman, Barbara. 1988. Employment aria WelfareIm acts of the Arkansas Work Pro ram: A Three-Year Follow-U Stud inTwo Counties.
CALIKANIA
Goldman, Barbara; Gueron, Judith; Ball, Joseph; Price, Marilyn; withFriedlander, Daniel; and Hamilton, Gayle. 1984. 2rellminary Findingsfrom the San Diego Job Search and Work Experience Demonstration.
Goldman, Barbara; Friedlander, Daniel; Gueron Judith; tong, David; withHamilton, Gayle; and Hoerz, Gregory. 1985. Findings from the SanDiego Job Search and Work Experience Demonstration.
Goldman, Barbara; Friedlander, Daniel; Long, David; with Erickson,Marjorie; and Gueron, Judith. 1986. Final Report on the San DiegoJob Search and Work Experience Demonstration.
-300- 3G 0
ILLINOIS
Manpower Demonst, ation Research Corporation. 1985. Baselire Paper on theEvaluation of the WIN Demonstration Program in Cook County _Illinois.
Quint, Janet; Guy, Cynthia; with Hoerz, Gregory; Hamilton, Gayle; Ball,Joseph; Goldman, Barbara; and Gueron, Judith 1986. Interim Findings
from the Illinois WIN Demonstration Progrm in Cook County.
Friedlander, Daniel; Freedman, Stephen; Hamilton, Gayle; Quint, Janet; withGoldman, Barbara; Long, David; and Ricci°, James. 1987. Final Report
on Job Search and Work Experience in Cook County.
MAINE
Auspos, Patricia; with Ball, Joseph; Goldman, Barbara; and Gueron, Judith.
1985. Maine: Interim Findings from a Grant Diversion Program.
Auspos, Patricia; Cave, George; Long, David; with Hanson, Karla; Caspar,Emma; Friedlander, Daniel; and Goldman, Barbara. 1988. Final Report
on the Training Opportunities in the PrivaLe Sector Program.
MARYLAND
Quint, Janet; with Ball, Joseph; Goldman, Barbara; Gueron, Judith; andHamilton, Gayle. 1984. Interim Findings from the Maryland Employment
Initiatives Programs.
Friedlander, Daniel; Hoerz, Gregory; Long, David; Quint, Janet; with
Goldman, arbara; and Gueron Judith. 1985. Maryland: Final Report on
the Empli.jment Initiatives Evaluaticm.
Friedlander, Daniel. 1787. Supplemental Report on the Baltimore OptionsProgram.
NEW JERSEY
Final Report, 1988.
VIRGIaIA
Price, Marilyn; wick Ball, Joseph; Goldman, Barbara; Gruber, David; Gueron,Judith; and Hamilton, Gayle. 1985. Interim Findings from theVirginia Employment Services Program.
361
-301-
Riccio, James; Cave, George; Freedman, Stephen; Price, Marilyn; withFriedlander, Daniel; Goldman, Barbara; Gueron, Judith; and Long,David. 1986. Final Report on the Virginia Employment Serv'.cesProgram.
WEST VIRGINIA
Ball, Joseph; with Hamilton, Gayle; Hoerz, Gregory; Goldman, Barbara; andGueron, Judith. 1984. West Virginia: Interim Findings on theCommunity Work Experience Demonstrations.
Friedlander, Daniel; Erickson, Marjorie; Hamilton, Gayle; Knox, Virginia;with Goldman, Barbara; Gueron, Judith; and Long, David. 1986. WestVirginia: Final Report on the Community Work Experience Demonstra-tions.
WEL_ARE GRANT DIVERSION
Bangser, Michael; Healy, James; and Ivry, Robert. 1985. Welfare GrantDiversion: Early Observations from Programs in Six States.
Bangser, Michael; Healy, James; and Ivry, Robert. 1986. Welfare GrantDiversion: Lessons and Prospects.
BROCHURE
Manpower Demonstration Research Corporation. 1987. 'Findings on StateWelfare Employment Programs.'
OTHER
Friedlander, Daniel; and Long, David. 1987. A Study of PerformanceMeasures and Subgroup Impacts in Three Welfare Employment Programs.
Friedlander, Daniel. 1988. Subgroup Impacts and Performance Indicators forSelected Welfare Employment Programs.
HaraUton, Gayle; with Ortiz, Vilma; Goldman, Barbara; Kierstead, C. Rudd;and Taylor, Electra. 1988. Interim Report on the Saturation WorkInitiative Model in San Diego.
Goldman, Barbara; Cavin, Edward; Erickson, Marjorie; Hamilton, Gayle;Hasselbring, Darlene; and Reynolds, Sandra. 1985. RelationshipBetween Earnin s and Welfare Benefits for Workin Reci ients: FourArea Case Studies.